Electric Vehicles Charging Sessions Classification
Technique for Optimized Battery Charge Based on Machine
Learning
Submitted by
LOKESH KUMAR V
(411420104030)
NAVEEN V
(411420104034)
in the partial fulfillment for the award of the degree
of
BACHELOR OF ENGINEERING
IN
COMPUTER SCIENCE AND ENGINEERING
NEW PRINCE SHRI BHAVANI COLLEGE OF
ENGINEERING AND TECHNOLOGY
ANNA UNIVERSITY:: CHENNAI 600 025
MAY 2024
ANNA UNIVERSITY :: CHENNAI 600 025
BONAFIDE CERTIFICATE
Certified that this project report “Electric Vehicles Charging Sessions Classification
Technique for Optimized Battery Charge Based on Machine Learning” is the bonafide
work of “LOKESH KUMAR V (411420104030), NAVEEN V (411420104034) ,
KANAGASELVI R (411420104023) ” who carried out the
project work under my supervision.
SIGNATURE
SIGNATURE
Dr.P.B.EDWIN PRABHAKAR
MS.P.KAVITHA M.E
HEAD OF THE DEPARTMENT
ASSISTANT PROFESSOR
Department of Computer Science
andEngineering
New Prince Shri Bhavani College of
Engineeringand Technology, Gowrivakkam,
Chennai-73.
Department of Computer Science and
Engineering
New Prince Shri Bhavani College of
Engineeringand Technology, Gowrivakkam,
Chennai-73.
Submitted for the Project Viva-voce Examination held on ………………
INTERNAL EXAMINER
EXTERNAL EXAMINER
ACKNOWLEDGEMENT
It gives me great pleasure to thank MS.P.KAVITHA, Professor, Department of
Computer Science and Engineering, for the constant support and guidance given to us
throughout the course of this project. She has been a constant source of inspiration for
us. We also take the opportunity to acknowledge the contribution of Dr.P.B.EDWIN
PRABHAKAR, Professor and Head of the Department, Department of Computer
Science and Engineering, for his support and assistance during the development of the
project. We also take this the opportunity to acknowledge the contribution of all faculty
members of the department for their assistance and cooperation during the development
of our project. We also thank all the Non-Teaching Staff of the Department who helped
us in the course of the project. Last but not the least, we acknowledge our friends for
their encouragement in the completionof the project.
1.
LOKESH KUMAR V (411420104005)
-
2.
NAVEEN V
-
(411420104034)
VISION
NPSBCET commits to strive for excellence in imparting technical education by
promoting innovation, creativity and entrepreneurial abilities of the students.
MISSION
1.
To enhance the effectiveness of teaching-learning process by providing a
stimulating learning environment.
2.
To establish R&D centers, incubation centers, centers of excellence in latest
technologies and provide a platform for students to interact with the industry.
3.
To achieve academic excellence by imparting knowledge and skills through
problem solving, practical training and design & development of innovative
projects.
4.
To sensitize students to social and environmental issues.
5.
To inculcate discipline in students and make them technologically and ethically
strong.
DEPARTMENT OF COMPUTERSCIENCE AND
ENGINEERING
VISION
To foster competent professional with ethical codes and make them Technologically
adept, self-motivated, and socially responsible innovators.
MISSION
M1:To stimulate challenging professional by imparting proficient education and the
zestof higher studies.
M2:Provide learning ambience to generate innovative and problem-solving skills with
professionalism.
M3:To imbibe leadership quality, thereby making them expertise in their career.
M4:To inculcate independent and lifelong learning with ethical and social responsibilities.
M5:To prepare highly qualified, sought-after, and technical intelligent strategists who can
expand the effectiveness.
PROGRAMME EDUCATIONAL OBJECTIVES (PEOS)
PEO1: To provide graduating students with core competencies by strengthening their
mathematical, scientific, and engineering fundamentals thereby pursue higher education
and research or have a successful career in industries associated with Computer Science
and Engineering, or as entrepreneurs.
PEO2: To train graduates in diversified and applied areas with analysis, design, and
synthesis of data to create novel products and solutions to meet current industrial and
societal needs.
PEO3:To
promote
collaborative
learning
and
spirit
of
teamwork
through
multidisciplinary projects and diverse professional activities. Also, inculcate high
professionalism amongthe students by providing technical and soft skills with ethical
standards.
PROGRAM SPECIFIC OBJECTIVES (PSOS)
1.
To analyze, design and develop computing solutions by applying foundational
concepts of Computer Science and Engineering.
2.
To apply software engineering principles and practices for developing quality
software for scientific and business applications.
COURSE OUTCOMES (COs)
1.
Identify technically and economically feasible problems of social relevance
2.
Plan and build the project team with assigned responsibilities .
3.
Identify and survey the relevant literature for getting exposed to related solutions
4.
Analyse, design and develop adaptable and reusable solutions of
minimalcomplexity by using modern tools .
5.
Implement and test solutions to trace against the user requirements
6.
Deploy and support the solutions for better manageability of the solutions and
provide scope for improvability.
PROGRAM OUTCOMES
Engineering Graduates will be able to:
1. Engineering knowledge: Apply the knowledge of mathematics, science,
engineering fundamentals, and an engineering specialization to the solution of
complex engineering problems.
2. Problem analysis: Identify, formulate, review research literature, and analyze
complex engineering problems reaching substantiated conclusions using first
principles of mathematics, natural sciences, and engineering sciences.
3. Design/development of solutions: Design solutions for complex engineering
problems and design system components or processes that meet the specified
needswith appropriate consideration for the public health and safety, and the cultural,
societal, and environmental considerations.
4. Conduct investigations of complex problems: Use research-based knowledge
and research methods including design of experiments, analysis and interpretation of
data, and synthesis of the information to provide valid conclusions.
5. Modern tool usage: Create, select, and apply appropriate techniques, resources,
and modern engineering and IT tools including prediction and modeling to complex
engineering activities with an understanding of the limitations.
6. The engineer and society: Apply reasoning informed by the contextual knowledge
to assess societal, health, safety, legal and cultural issues and the consequent
responsibilities relevant to the professional engineering practice.
7. Environment and sustainability: Understand the impact of the professional
engineering solutions in societal and environmental contexts, and demonstrate the
knowledge of, and need for sustainable development.
8. Ethics: Apply ethical principles and commit to professional ethics and
responsibilities and norms of the engineering practice.
9. Individual and team work: Function effectively as an individual, and as a member
or leader in diverse teams, and in multidisciplinary settings.
10. Communication: Communicate effectively on complex engineering activities
with the engineering community and with society at large, such as, being able to
comprehend and write effective reports and design documentation, make effective
presentations, and give and receive clear instructions.
11. Project management and finance: Demonstrate knowledge and understanding
of the engineering and management principles and apply these to one’s own work, as
a member and leader in a team, to manage projects and in multidisciplinary
environments.
12. Life-long learning: Recognize the need for, and have the preparation and ability
to engage in independent and life-long learning in the broadest context of
technological change.
ABSTRACT
Electric-powered vehicles will help reduce greenhouse gas emissions and increase fuel
prices. The main purpose of wireless transmission in electric vehicles is to transfer
power over a small distance. The wireless power transmission system consists of a
transmitter and receiver part that is separated by a small distance. Wireless transmission
technology uses a flexible electromagnetic field. This electric field is created in a free
environment that carries a fixed amount of money that creates a magnetic field around it
and this field contains energy in it and the EMF is generated between the coils and
transmitted to the receiver. BMS is a battery management system. In EV vehicles we use
two batteries such as master and slave. The first preference is given to the master battery
in BMS. If the master battery charge comes down automatically the relay will switch
from master battery to slave battery.
IX
திட்டம் சுறுக்கம் :
மின்சாரத்தில்
இயங் கும்
வவளியயற் றத்ததக்
அதிகரிக்கவும்
சக்திதய
குதறக்கவும் ,
உதவும் .
டிரான்ஸ்மிஷனின்
வாகனங் கள்
மாற் றுவதாகும் .
சிஸ்டம்
ஒரு
எரிவ ாருள்
மின்சார
முக்கிய
கிரீன்ஹவுஸ்
ஒரு
வயர்வலஸ்
டிரான்ஸ்மிட்டர்
விதலதய
வாகனங் களில்
ய ாக்கம்
மற் றும்
வயர்வலஸ்
சிறிய
வர்
வாயு
தூரத்திற் கு
டிரான்ஸ்மிஷன்
ரிசீவர்
குதிதயக்
வகாண்டுள் ளது, இது ஒரு சிறிய தூரத்தால் பிரிக்க ் ட்டுள் ளது.
வயர்வலஸ்
டிரான்ஸ்மிஷன்
மின்கா ்த புலத்தத ்
வதாழில் நுட் ம்
வ கிழ் வான
யன் டுத்துகிறது. இ ்த மின்சார புலம் ஒரு
இலவச சூழலில் உருவாக்க ் ட்டது, அது ஒரு
வகாண்டு
ஒரு
வசல் கிறது,
அது
சுற் றி
ிதலயான
ஒரு
ணத்தத
கா ்த ் புலத்தத
உருவாக்குகிறது மற் றும் இ ் த புலத்தில் ஆற் றல் உள் ளது மற் றும்
EMF
சுருள் களுக்கு
அனு ்
இதடயில்
உருவாக்க ் ட்டு
வ று ருக்கு
் டுகிறது. BMS என் து ய ட்டரி யமலாண்தம அதம ் பு. EV
வாகனங் களில்
மாஸ்டர்
ய ட்டரிகதள ்
ய ட்டரிக்கு
முதல்
மற் றும்
என
இரண்டு
பிஎம் எஸ்ஸில்
மாஸ்டர்
அளிக்க ் டுகிறது.
மாஸ்டர்
யன் டுத்துகியறாம் .
முன்னுரிதம
X
ஸ்யலவ்
ய ட்டரி சார்ஜ் தானாகயவ குதற ்துவிட்டால் , ரியல முதன்தம
ய ட்டரியிலிரு ்து ஸ்யலவ் ய ட்டரிக்கு மாறும் .
.
XI
TABLE OF CONTENTS
CHAPTER NO.
TITLE
ABSTRACT
IX
XVI
LIST OF FIGURES
LIST OF ABBREVIATIONS
1.
PAGE NO.
XVIII
INTRODUCTION
1
1.1 PROBLEM STATEMENT
1
1.2 OBJECTIVES
2
1.3 SCOPE
3
2.
LITERATURE REVIEW
4
3.
EXISTING SYSTEM
13
4.
PROPOSED SYSTEM
16
4.1 INTRODUCTION
16
4.1.1 Support Vector Machine
16
4.1.2 Microservices Architecture
16
4.1.3 Cloud Computing
17
4.1.4 Data Security Measures
17
4.1.5 Objective Evaluation
18
XII
4.2 SOFTWARE AND HARDWARE SPECIFICATIONS
18
4.2.1 Software Requirements
18
4.2.2 Hardware Requirements
18
4.3 SOFTWARE DESCRIPTION
19
4.4 PROJECT OVERVIEW
22
4.5 MATLAB OVERVIEW
22
4.6 MATLAB SYSTEM
24
4.7 MATLAB DOCUMENTATION
25
4.8 MATLAB ONLINE HELP
26
4.9 MATLAB’S POWER OF COMPUTATION
27
4.10 HIGHLIGHTS OF MATLAB
28
4.11 EMPLOYMENTS OF MATLAB
29
4.12 ENVIRONMENT SETUP
29
4.12.1 Neighborhood Setup
29
4.12.2 Understanding MATLAB Environment
35
4.13 MAIN PARTS OF GUI
42
4.13.1 Common Information
42
4.13.2 Buttons and Sliders
45
4.13.3 Self-Organizing Map
53
4.13.4 Performance
54
XIII
4.13.5 Use of SOM toolbar
5.
6.
57
4.14 CONSTRUCTION OF DATASETS
58
4.15 DATA PREPROCESSING
59
4.16 INITIALIZATION AND TRAINING
61
CODING AND OUTPUT
69
5.1 CODE
69
5.2 OUTPUT SCREEN SHOT
76
RESULT AND DISCUSSION
79
6.1 EXPERIMENTAL SETUP
79
6.2 EFFICIENCY OF ROUTE PLANNING ALGORITHMS
80
6.3 SCALABILITY OF MICROSERVICES ARCHITECTURE 80
6.4 ROBUSTNESS OF DATA SECURITY MEASURES
81
6.5 DISCUSSION
81
7.
CONCLUSION
83
8.
FUTURE ENHANCEMENTS
86
7.1 CO-PO-PSO MAPPING
89
7.2 SUBJECT MAPPING
89
9.
REFERENCE
90
XIV
LIST OF FIGURES
FIG NO
NAME OF THE FIGURE
PAGE NO
4.1
Data Mining Tools Poll
20
4.2
File Installation Key
30
4.3
Mathworks Installer
30
4.4
Folder Selection
30
4.5
License Agreement
31
4.6
Product Selection
31
4.7
License File
32
4.8
Installation Options
32
4.9
Confirmation
33
4.10
Product Configuration Notes
33
4.11
Installation
34
4.12
Complete Installation
34
4.13
Mathwork Account Creation
35
4.14
Mathlab IDE
35
4.15
Current Folder
36
4.16
Order Window
36
XV
4.17
Workspace Window
37
4.18
Diagram of Project Chapter
37
XVI
4.19
Case of Geographical UI With a Portion of the Segments
41
4.20
Property Inspector
43
4.21
An Example of Property Inspector for a Slider Bar Axes
47
4.22
An Example of Property Inspector for Axes Creating Menu
48
4.23
An Exemplary Menu Created in Menu Editor
49
4.24
Simple GUI with Ready Built Menu
51
4.25
Global Mapsheet
53
4.26
SOM Toolbox
54
4.27
Dataset Prepositioning Tool
61
4.28
SOM Initialization and Training Tool
61
4.29
Visualization of the SOM of IRIS Data
66
4.30
Projection of the IRIS Dataset
67
XVI
LIST OF ABBREVIATIONS
S NO
ABBREVIATION
EXPANSION
1.
SVM
Support Vector Machine
2.
GIS
Geographic Information System
3.
HILF
High- Impact Low Frequency
4.
PMESOPM
Proactive Mobile Energy Storage
Optimal Model
5.
FRR
Flood Recovery Rate
6.
PCA
Principal Component Analysis
7.
POI
Point Of Interest
8.
SOM
Self Organizing Map
9.
BMU
Best Matching Unit
10.
sD
Som_denormalize
11.
sM
Some_Makes
12.
IOT
Internet Of Things
13.
ML
Machine Learning
14.
IES
International Electronics
Symposium
15.
ICPES
International Conference On Power
and Energy Systems
XVII
CHAPTER I
INTRODUCTION
Monsoon-induced flooding is a recurring natural disaster in many regions,
causing widespread destruction and posing significant challenges to rescue and relief
efforts. Traditional approaches to flood risk assessment and rescue operations often
rely on simplistic methods such as K-Means clustering, which may lack accuracy and
fail to adapt to the complexities of dynamic flood patterns. Moreover, the increasing
frequency and severity of floods due to climate change underscore the need for more
sophisticated and resilient frameworks to mitigate their impact.
Recent advancements in technology, particularly in the fields of machine
learning, cloud computing, and geographic information systems (GIS), offer new
opportunities to enhance flood rescue operations. These technologies enable the
development of more precise predictive models, efficient route planning algorithms,
and robust infrastructure for data storage and processing. By harnessing these
capabilities, it becomes possible to create a comprehensive and adaptable framework
capable of improving the accuracy and effectiveness of flood rescue operations.
1.1 Problem Statement
The existing methodologies for flood risk assessment and rescue operations are
often inadequate in accurately identifying flood-prone regions and optimizing rescue
1
routes, particularly in the context of monsoon-induced flooding. Traditional
clustering algorithms like K-Means may struggle to capture the complex spatial and
temporal dynamics of floods, leading to suboptimal outcomes in terms of rescue
operation efficiency and effectiveness.
Furthermore, the lack of integration between disparate systems and the absence
of real-time data processing capabilities hinder the timely response to unfolding flood
situations, potentially exacerbating the impact on affected communities. Addressing
these challenges requires the development of a more advanced and integrated
framework that leverages cutting-edge technologies to improve flood risk prediction,
route planning, and overall disaster resilience.
1.2 Objectives
The primary objective of this research is to develop an enhanced framework
for flood rescue operations that leverages advanced technologies such as machine
learning, cloud computing, and GIS mapping. Specific objectives include:
1. Implementing a Support Vector Machine (SVM) algorithm for more accurate and
reliable flood risk prediction.
2. Integrating microservices architecture to enhance system scalability, flexibility,
and resilience.
3. Developing optimized route planning algorithms using techniques such as hybrid
A to improve the efficiency of rescue operations.
4. Establishing robust data security measures to safeguard sensitive information and
2
ensure compliance with privacy regulations.
5. Validating the proposed framework through rigorous testing and evaluation,
comparing its performance against existing methodologies.
1.3 Scope
The proposed framework focuses specifically on addressing the challenges associated
with monsoon-induced flooding and the corresponding rescue operations. Key
aspects within the scope of this research include:
1. Flood risk assessment: Developing predictive models to identify flood-prone
regions with higher accuracy and precision.
2. Route planning: Optimizing rescue routes using advanced algorithms to minimize
response time and maximize resource utilization.
3. Technology integration: Leveraging microservices architecture to create a scalable
and adaptable framework capable of handling diverse data sources and processing
requirements.
4. Security and compliance: Implementing robust data security measures to protect
sensitive information and ensure regulatory compliance.
5. Validation and evaluation: Conducting comprehensive testing and validation
procedures to assess the performance and effectiveness of the proposed framework
under various scenarios and conditions.
3
CHAPTER II
LITERATURE REVIEW
In 2023 R. R. Sarkar, M. N. Islam, R. Islam, M. S. Hasan and M. Zahidur
Rahman presented Empowering Resilience in Post-Disaster Communication
with Low-End Communication Devices
Natural or man-made disasters result in significant loss of life, property damage,
disruption of communication networks, and immense suffering for survivors. When
a disaster strikes, the communication infrastructure is frequently rendered unreliable
or completely inoperable. In such cases, it is critical to establish emergency
communication to lessen the aftermath of the disaster, rescue those in need, and
accelerate relief efforts. This paper presents a communication model to empower
resilience in communication after a disaster and also relief and rescue efforts. This
model divides communication strategies into two categories namely communication
between victims and rescuers (V2R) and communication among rescuers (R2R). This
model collects victims' urgent aid messages, allowing rescuers to respond quickly and
increase the speed of rescue and relief operations. Rescuers in the second strategy are
able to communicate with one another. When combined, these two strategies form the
model, ultimately leading to the acceleration of comprehensive rescue and relief
operations and the mitigation of post-disaster consequences.
4
In 2022 W. Wang, S. Gao, H. Zhang, D. Li and L. Fu presented Resilience
Assessment and Enhancement Strategies of Transmission System under
Extreme Ice Disaster
Ice storm event with high impact and low probability causes a huge challenge to the
normal operation of the transmission system. To assess and enhance the resilience of
the transmission system under an ice disaster, this paper constructs a resilience
assessment and enhancement method for the transmission system. Firstly, the failure
rate model of the transmission line is established according to the characteristics of
the ice disaster scenario. Then, the resilience assessment metrics are constructed by
analyzing the whole process of the system resilience under an ice disaster. On this
basis, a resilience enhancement method under the ice disaster is proposed by using
the transfer entropy of power flow to screen the lines that need deicing. Finally, the
IEEE-30 bus transmission system is utilized to assess the resilience of the
transmission system and verify the effectiveness of the proposed resilience
enhancement method.
In 2023 Y. Zhang, L. Xu, C. Deng, W. Mao, H. Jiang and L. Li presented
Resilience Improvement Strategy of Distribution Network Based on Network
Reconfiguration in Earthquake Disaster Scenario
In recent years, the occurrence of extreme disaster events has caused serious impact
on the stable and reliable operation of the power system. Although the probability of
such events is small, they often bring great harm. Therefore, it is very necessary to
5
improve the resilience of power system to deal with such small probability and highrisk disaster events. In this paper, the resilience evaluation framework of transmission
and distribution system under earthquake disaster is put forward, and the resilience
index of power system is measured by the shearing load of the system under fault
state. The probability and degree of fault of transmission and distribution system
under different earthquake intensity are simulated by earthquake disaster model.
Taking IEEE 33 model as an example, three typical fault scenarios are selected, and
the optimal load reduction is obtained by using contact switches to reconstruct the
distribution network according to different fault locations. The example results show
that the proposed method can significantly improve the resilience of the distribution
network under earthquake disasters, which provides a reference for improving the
resilience performance of the power system under earthquake disasters.
In 2023 J. Yuan, C. Wan, J. Huang and T. Wang presented Developing Risk
Reduction Strategies of Typhoon Disaster for Ports from the Perspective of
Resilience
With the deepening of economic globalization, the strategic significance of ports is
rising. However, ports face various risks and challenges from both internal and
external sources. This paper introduces the resilience theory into the port safety risk
management, and explores the port resilience change and the corresponding risk
response mechanism effect when facing typhoon disaster. The definition and
characteristics of port safety resilience are analyzed, and a triangular model of port
6
safety resilience is constructed. Taking the container supply capacity as the
performance index, using the functional level function to generate the port system
resilience curve, according to the curve change characteristics, the safety resilience
of the port system is analyzed and evaluated in four stages. Combined with the basic
conditions of the target port and the operating data of mechanical equipment, the
change of the safety resilience of the port logistics risk system after the typhoon attack
was simulated. The joint risk coping strategies were developed from the aspects of
single port and port cluster respectively, and the effects of different strategies on port
safety resilience were evaluated under the influence of typhoon disaster. This paper
applies the tenacity theory to the research of port system safety management, and
realizes the quantitative evaluation and assessment of port tenacity under typhoon
disaster by means of risk modeling and simulation, which provides reference for port
disaster prevention and mitigation.
In 2022 A. Younesi, Z. Wang and L. Wang presented Investigating the Impacts
of Climate Change and Natural Disasters on the Feasibility of Power System
Resilience
Due to the increasing rate of high-impact low-frequency (HILF) events, power
systems are more vulnerable against the destructive climate events compared to other
infrastructures. From this point of view, the primary focus of this article is to
investigate the vulnerability of power systems in the face of numerous types of natural
disasters in terms of resilience metrics. To achieve this goal, a mesh-structured view
7
of the power system at the transmission level is employed to model the action
mechanism from different types of natural disasters on the power system. The Monte
Carlo simulation method is further applied to evaluate the resilience metrics of the
power system. From the perspective of resilience, the vulnerability of the system
against different types of events is finally achieved in this paper. Simulation case
studies on the IEEE 30-bus test system have demonstrated that the proposed modeling
can not only facilitate in upgraded schemes, but also significantly decrease the amount
of damages to the power system after natural extreme events.
In 2023 S. Aghababaei, M. T. Kenari, M. S. Sepasian and A. Ozdemir presented
Proactive allocation of mobile energy storage systems before a natural disaster
to improve distribution system resilience
In the last few years, the vulnerability of distribution systems against extreme
catastrophes has led electric companies to move towards resilient networks.
Meanwhile, battery energy storage systems in distribution grids have been considered
a promising solution due to their technical and economic advantages. This study
proposes a proactive mobile energy storage optimal placement model (PMESOPM)
to enhance the resilience of the power distribution system before a natural disaster. In
this model, according to the fragility curve and probability of failure of components
for lines and roads, Monte Carlo simulation is used to identify the failure states of any
component in each iteration. Then, a pre-hurricane approach is adopted using the
combination of genetic and Floyd algorithms to deploy mobile storage systems the
8
day before the storm or hurricane. The numerical analysis is carried out using the
IEEE 33-bus standard test network, mapped on the Sioux Falls traffic network. The
results validate the effectiveness of the proposed model in critical conditions of the
network.
In 2022 Y. Yang, W. Lili and Z. Hongchi presented Bibliometric research on the
evolution of resilience theme from the perspective of Geographical Science
Resilience is trying to influence regional, city, and village planning, construction, and
development. This research analyzes the thematic evolution of regional resilience,
urban resilience, and rural resilience in terms of the number of documents published,
keywords co-occurrence, clustering, and burst. Regional economic resilience has
been identified as one of the key contents of regional resilience research. The main
contents of resilient city research are concept definition, construction strategy,
resilient city planning, and resilient governance. Rural resilience is considered as a
new field of research. Improving rural resilience makes a significant contribution to
disaster prevention and reduction, as well as closing the gap between rich and poor.
Cross-field linkage research should be done in the future, and research on the
mechanism of the resilience process should be strengthened.
In 2022 S. Kim and Y. -W. Kwon presented Construction of Disaster Knowledge
Graphs to Enhance Disaster Resilience
As a result of the recent surge in disaster-related data, numerous studies have been
9
conducted to deal with the massive amount of data. In the meantime, the issue of
managing data in various formats and representing their relevance is being raised. In
this paper, we present a disaster knowledge graph to analyze the impact of a disaster
and predict how much effort it will take to recover from the disaster. To that end, we
define the structure of a disaster knowledge graph containing data collected from
sensors, social networks, web, and risk analysis results. To extract meaningful
information from structured and unstructured data, we use a risk analysis platform
that can compute hazard values in accordance with various hazard models. Then, we
store automatically graphs into a graph database as a form of a time-series data.
Therefore, it will be possible to predict the progress of a complex disaster that can
occur in a chain using a series of disaster knowledge graphs.
In 2023 L. Yi et al. presented Distributionally Robust Resilience Enhancement
Model for the Power Distribution System Considering the Uncertainty of
Natural Disasters
Natural disasters with high risk and lower occurrence probability have attracted much
more concern in recent years. In this paper, we proposed a distributionally robust
resilience enhancement model for the distribution power system, in which the
uncertainties of natural disasters are also taken into consideration. The ambiguity of
the DRRM is constructed based on the branch outage probability, and the nested CCG
algorithm is applied to solve the proposed model. The DRRM has been verified in the
IEEE 33-bus distribution system. Case studies showed that the proposed model can
10
reach a more effective and economic reinforcement strategy for the power distribution
system.
In 2023 T. Zheng, F. Wu, C. Wang and L. Lu presented Assessing Urban
Resilience to Flooding at County Level Using Multi-Modal Geospatial Data
Urban resilience refers to the capacity of an urban system to adapt and respond to
changes, including the ability to better cope with future disaster risks. With the
intensifying impact of global climate change, cities are becoming more vulnerable to
natural disasters. It is crucial for cities to effectively resist and maintain sustainable
economic and social development in the face of these disasters. This paper, taking the
“2021.07.20 Henan rainstorm” flood disaster in the Weihe river basin as a study case,
applying Sentinel-1 (S1) synthetic aperture radar (SAR) images and other multimodal geospatial data, aims to assess county-scale urban resilience against flooding.
First, the random forest classifier was adopted to extract water bodies at periods of
pre-flood, during-flood and post-flood from the preprocessed S1 data. Second, the
flood recovery rate (FRR) was chosen for representing urban flood resilience, and
was calculated at county-level based on the water bodies of the three periods. Third,
data of the 12 factors of social, economic, community and environment dimensions
were collected and transformed, and were used to explore and evaluate the main
impacting factors on county-level FRRs with the aid of Pearson correlation analysis
and principal component analysis (PCA). The results show that: 1) Districts in the
southwest have higher recovery levels, while districts in the east have lower recovery
11
levels. 2) The four factors of points of interest (POI) all have significant positive
effects on FRR, while topography and slope have considerable negative impacts on
FRR. 3) The distribution of FRR and the weights of factors’ influence on FRR can be
combined for developing relevant policies for enhancing urban flood resilience.
12
CHAPTER III
EXISTING SYSTEM
The existing system for addressing monsoon-induced flooding and conducting
rescue operations revolves around traditional methodologies that often struggle to
cope with the complexities of dynamic flood patterns and changing environmental
conditions. These methodologies typically rely on simplistic approaches such as KMeans clustering for flood risk assessment and route planning. While these methods
have been utilized for some time, their limitations become increasingly apparent as
the frequency and severity of floods escalate due to factors such as climate change.
One of the primary challenges with the existing system is its inability to
accurately identify flood-prone regions and predict the extent of flooding with
sufficient precision. K-Means clustering, for instance, partitions data into clusters
based on similarity, often leading to oversimplified representations of flood patterns
and inadequate risk assessments. As a result, rescue operations may be inefficiently
allocated or fail to reach areas most in need of assistance in a timely manner,
exacerbating the impact on affected communities.
Moreover, the existing system often lacks integration between disparate data
sources and platforms, hindering the seamless exchange of information and real-time
decision-making during flood events. Without the ability to access and analyze data
13
in a timely manner, responders may struggle to coordinate efforts effectively, leading
to delays in rescue operations and potentially increasing the risk to both responders
and affected populations.
Another significant challenge is the limited scalability and adaptability of the
existing system, particularly in the face of evolving technological advancements and
changing environmental conditions. Traditional methodologies may struggle to
incorporate new data sources or adapt to emerging trends, resulting in outdated and
inefficient approaches to flood risk assessment and rescue operations. Additionally,
the lack of robust data security measures may expose sensitive information to
unauthorized access or compromise, posing further risks to the integrity and reliability
of the system.
Overall, the existing system for addressing monsoon-induced flooding and
conducting rescue operations is characterized by its reliance on outdated
methodologies, limited integration between disparate platforms, and inadequate
scalability and adaptability to evolving challenges. To address these shortcomings
and improve the effectiveness of flood response efforts, there is a pressing need for
the development of a more advanced and integrated framework that leverages cuttingedge technologies such as machine learning, cloud computing, and geographic
information systems (GIS). Such a framework would enable more accurate flood risk
assessment, efficient route planning, and secure data management, ultimately
14
enhancing the resilience and effectiveness of flood rescue operations.
Furthermore, the existing system often lacks the capability to incorporate realtime data feeds from various sensors and monitoring devices, limiting its ability to
provide up-to-date situational awareness during flood events. This deficiency can
impede decision-making processes and hinder the coordination of rescue efforts,
potentially resulting in suboptimal outcomes and increased risks to both responders
and affected communities. Additionally, the lack of interoperability between different
systems and platforms can lead to data silos and fragmentation, further complicating
the sharing and analysis of critical information. Overall, the existing system's
shortcomings highlight the urgent need for a more advanced and integrated approach
to flood risk assessment and rescue operations that can effectively address the
challenges posed by monsoon-induced flooding and enhance overall disaster
resilience.
15
CHAPTER IV
PROPOSED SYSTEM
4.1 Introduction
The proposed system aims to revolutionize flood rescue operations by
introducing a comprehensive and technologically advanced framework that leverages
cutting-edge techniques such as the Support Vector Machine (SVM) algorithm,
microservices architecture, cloud computing, and robust data security measures. This
section provides an overview of the proposed system and outlines its key components
and functionalities.
4.1.1 Support Vector Machine (SVM) Algorithm
The heart of the proposed system lies in the adoption of the Support Vector
Machine (SVM) algorithm for flood risk prediction. Unlike traditional clustering
algorithms such as K-Means, SVM offers superior capabilities in handling complex
data patterns and achieving higher accuracy in predictive modeling. By leveraging
SVM, the proposed system aims to improve the precision and reliability of flood risk
assessment, enabling more effective allocation of resources and timely response to
flood events.
4.1.2 Microservices Architecture
In addition to utilizing advanced machine learning techniques, the proposed
16
system embraces a microservices architecture to enhance scalability, flexibility, and
resilience. Microservices break down complex systems into smaller, independently
deployable units, allowing for easier integration of new functionalities and seamless
adaptation to changing requirements. By adopting a microservices-based approach,
the proposed system can efficiently manage various aspects of flood rescue
operations, including fleet management, route planning, data processing, and
communication.
4.1.3 Cloud Computing
Cloud computing plays a crucial role in the proposed system by providing
scalable and on-demand access to computing resources, storage, and services. By
leveraging cloud infrastructure, the system can handle large volumes of data, perform
complex computational tasks, and support real-time decision-making processes
during flood events. Moreover, cloud-based solutions offer increased flexibility and
cost-effectiveness compared to traditional on-premises infrastructure, making them
well-suited for dynamic and resource-intensive applications like flood rescue
operations.
4.1.4 Data Security Measures
Ensuring the security and integrity of sensitive information is paramount in any
disaster resilience framework. To address this concern, the proposed system
incorporates robust data security measures to safeguard against unauthorized access,
17
data breaches, and cyber threats. This includes encryption, access control
mechanisms, intrusion detection systems, and regular security audits to identify and
mitigate potential vulnerabilities. By prioritizing data security, the proposed system
aims to build trust and confidence among stakeholders and mitigate the risks
associated with handling sensitive information in emergency situations.
4.1.5 Objective Evaluation
The proposed system's effectiveness will be rigorously evaluated against
predefined objectives and performance metrics. This includes assessing the
accuracy of flood risk prediction, the efficiency of route planning algorithms,
the scalability of microservices architecture, and the robustness of data security
measures. Real-world deployment scenarios and simulated flood events will be
used to validate the system's capabilities and identify areas for improvement.
Additionally, user feedback and stakeholder input will be solicited to ensure
the proposed system meets the needs and expectations of end-users and
decision-makers involved in flood rescue operations.
4.2 Software & Hardware Specifications
4.2.1 Software Requirements:
 Tool: Matlab
 Language: Python
4.2.2 Hardware Requirements:
18

Hard Disk: Greater than 500 GB

RAM: Greater than 4 GB
 Processor: Core 2 Duo and Above
4.3 Software Description:
MATLAB is a great and flexible tool, more than accomplish of
performing the data mining. It is clear that MATLAB has not to be given due
concentration in this arena. Figure 4.1 illustrate the, while a comparatively
trendy data mining tool, MATLAB is not so far in the group of packages
such as Clementine, Weka and still Excel. In adding together, though
MATLAB is selected more regularly than Oracle, it is usually used in
combination with other tools. Where-as Oracle is implementing as the standalone tool over 50% of the time, MATLAB is use on its own just over a 12%
of the time.
Summarises the place of MATLAB over the last past 7 years. in spite
of the fact that MATLAB is presently capable of the stage, some of the most
trendy data mining technique existing, such as those being analyse this
project, it has not so far become one of the groups of choice in this meadow.
The popularity of these methods is detailed in Table 4.1, which is based on
a samples of 16 altered data mining methodsover the last 4 year period from
2013 to 2016.
19
Figure 4.1 2016 Data Mining Tools Poll 1138 Votes
MATLAB Ranks 10th with 5% of the votes
One causes for MATLAB’s restricted use may be the fact that is a proprietary group
(or) package. However, the fundamental MATLAB package is without difficulty
enhanced, mainly by using the open-source tool-boxes and the script bundles, such
as those examine in this case study. The detail MATLAB’s data mining possible has
positively not been entirely subjugated (as established in Figure 4.1 and Table 4.2),
20
jointly with the current required for data mining tools, is the middle inspiration for
carrying out this case study.
Method
2013
2014
2015
2016
Rank:1
Rank:1
Rank:1
(15%)
(15%)
(16%)
(13%)
Rank:2
Rank:2
Rank:3
Rank:2
(11%)
(11%)
(10%)
(12%)
Rank:5
Rank:4
Rank:5
Rank:6
(8%)
(8%)
(8%)
(7%)
Association
Rank:6
Rank:7
Rank:4
Rank:7
rules
(7%)
(4%)
(8%)
(6%)
Decision tree Rank:1
Clustering
Neural nets
Table 4.1: Polls of trendy Data Mining Methods 2013-2016
MATLAB
2010
2011
2012
2013
2014
2015
2016
Rank
∞
7.0
7.0
14.0
9.0
15.0
10.0
5%
5%
3%
2%
2%
5%
Percentage N/a
Table 4.2: celebrity of MATLAB in Data Mining 2010-2016
The combination of data mining tools provide the thesis allowed for an far large
holistic technique to data mining in MATLAB than has been presented existing
and in the addition, ensure the MATLAB can be use as a stand-alone tool,
somewhat than in combination with former packages. These case studies ensure
that data mining in MATLAB become a gradually more clear-cut task, as the
21
suitable tools for a known investigation become visible. As a logical expansion of
the combination provide, recommendation is given with consider the formation of
a data mining toolbox for MATLAB. The opportunity for addition to this workis
numerous, not only in terms of extend the tools them-selves but andalso of data
mining in MATLAB as an entire.
4.4 Project Overview
Due to the broad and undefined environment of this case study it is very
important that we focal point on the number of exact tools and case study. The data
mining tools in the region of which this study case will revolve are: the NeuralNetwork Toolbox, a proprietary tool presented from The Math-Works, distributors
of MATLAB. The Fuzzy cluster and Data study Toolbox [Balasko et al. 2015] and
the association Rule Miner and presumption study tool [Malone 2013], which
are both open-platform; and lastly an execution of the C4.5 judgment tree
method [Woolf, 2015].
4.5 MATLAB Overview
MATLAB is an elite dialect for specialized figuring. It incorporates
calculation, perception, and programming in a simple to- utilize condition where
issues and arrangements are communicated in commonplace numerical
documentation.
Common place uses incorporate
22
 Math and calculation
 Algorithm improvement
 Data obtaining
 Modeling, re enancetment, and prototyping
 Data investigation, investigation, and representation
 Scientific and building illustrations
 Application improvement, including graphical UI building
MATLAB is an intelligent framework whose fundamental information
component is an exhibit that does not require dimensioning. This enables you to
explain numerous specialized figuring issues, particularly those with grid and
vector definitions, in a small amount of the time it would take to compose a
program in a scalar non interactive dialect, for example, C or Fortran.
The name MATLAB remains for lattice research facility. MATLAB was initially
written to give simple access to lattice programming created by the LINPACK
what's more, EISPACK ventures. Today, MATLAB motors join the LAPACK
what's more, BLAS libraries, installing the bestin class in programming for lattice
calculation.
MATLAB has developed over a time of years with contribution from numerous
clients. In college conditions, it is the standard instructional device for starting
what’s more, best in class courses in arithmetic, building, and science. In industry,
MATLAB is the device of decision for high-efficiency research, improvement, and
23
investigation.
MATLAB highlights a group of extra application-particulararrangements called
tool kits. Important to most clients of MATLAB, toolstash enables you to learn and
apply specific innovation. Tool kits are complete accumulations of MATLAB
capacities (M-records) that expand the MATLAB condition to take care of specific
classes of issues. Regions in which tool kits are accessible incorporate flag
handling, control frameworks, neural systems, fluffy rationale, wavelets, reenactment, and numerous others.
4.6 THE MATLAB SYSTEM
The MATLAB framework comprises of five fundamental parts:
 Improvement Environment. This is the arrangement of apparatuses and offices
that assistance you utilize MATLAB capacities and records. A considerable lot of
these instrumentsare graphical UIs. It incorporates the MATLAB work area and
Command Window, a charge history, an editorial managerand debugger, and
programs for review help, the workspace, records, what's more, the inquiry way.
 The MATLAB Mathematical Function Library. This is a huge gathering of
computational calculations going from basic capacities, similar to total, sine,
cosine, and complex number- crunching, to more advanced capacities like network
backwards, framework eigen values, Bessel capacities, and quick Fourier changes.
 The MATLAB Language. This is an abnormal state framework/exhibit dialect
with control stream proclamations, capacities, information structures, input/yield,
24
and protest situated programming highlights. It permits both "programming in the
little" to quickly make snappy discard projects, and "programming in the huge" to
make substantial and complex application programs.
 Designs. MATLAB has broad offices for showing vectors and lattices as
diagrams, and additionally commenting on and printing these charts. It
incorporates abnormal state capacities for two-dimensional and three-dimensional
information perception, picture handling, activity, and introduction illustrations. It
too incorporates low-level capacities that enable you to completely tweak the
presence of illustrations and in addition to assemble finish graphical UIs on your
MATLAB applications.
 The MATLAB External Interfaces/API. This is a librarythat enables you
to compose C and Fortran programs that collaborate with MATLAB. It
incorporates offices for calling schedules from MATLAB (dynamic connecting),
calling MATLAB as a computational motor, and for perusing and composing
MAT-records.
4.7 MATLAB DOCUMENTATION
MATLAB gives broad documentation, in both printed and on the web
design, to enable you to find out about and utilize the greater part of its highlights.
In the event that you are another client, begin with this Getting Started book. It
covers all the essential MATLAB highlights at an abnormal state, including
numerous cases.
25
The MATLAB online help gives undertaking focused and reference data about
MATLAB highlights. MATLAB documentation is additionally accessible in
printed shape and in PDF organizes.
4.8 MATLAB ONLINE HELP
To see the online documentation, select MATLAB Help from the Help menu in
MATLAB. The MATLAB documentation is sorted out into these principle themes:
 Desktop Tools and Development Environment — Startup and shutdown, the
work area, and different devices that assistance you utilize MATLAB
 Mathematics — Mathematical tasks and information investigation
 Programming — The MATLAB dialect and how to create MATLAB
applications
 Graphics — Tools and systems for plotting, diagram explanation, printing,
furthermore, programming with Handle Graphics®
 3-D
Visualization
—
Visualizing
surface
and
volume information,
straightforwardness, and review and lightingsystems
 Creating Graphical User Interfaces — GUI-building devices and how to
compose callback capacities
 External Interfaces/API — MEX-documents, the MATLAB motor, and
interfacing to Java, COM, and the serial port
MATLAB additionally incorporates reference documentation forall MATLAB
26
capacities:
 Functions - By Category — Lists all MATLAB capacities assembled into
classifications
 Handle Graphics Property Browser — Provides simple access to depictions of
designs protest properties
 External Interfaces/API Reference — Covers those capacities utilized by the
MATLAB outside interfaces, giving data on language structure in the calling
dialect, portrayal, contentions, return esteems, and illustrations
The MATLAB online documentation likewise incorporates
• Examples
—
A
record
of
cases
incorporated
into
the
documentation
• Release Notes — New highlights and known issues in the presentdischarge
• Printable Documentation — PDF forms of the documentationappropriate
for printing
4.9 MATLAB'S POWER OF COMPUTATIONAL MATHEMATICS
MATLAB is utilized as a part of each feature of computational science. Following
are a few regularly utilized scientific counts where it is utilizedgenerally usually:
 Dealing with Matrices and Arrays
 2-D and 3-D Plotting and illustrations
 Linear Algebra
 Algebraic Equations
27
 Non-straight Functions
 Statistics
 Data Analysis
 Calculus and Differential Equations
 Numerical Calculations
 Integration
 Transforms
 Curve Fitting
 Various other exceptional capacities
4.10 HIGHLIGHTS OF MATLAB
Following are the essential highlights of MATLAB:
 It is an abnormal state dialect for numerical calculation, representation and
application advancement.
 It additionally gives an intelligent domain to iterative investigation, plan what's
more, critical thinking.
 It gives immense library of numerical capacities for direct variable based math,
measurements, Fourier examination, sifting, advancement, numerical
coordination and comprehending standard differential conditions.
 It gives worked in illustrations to picturing information and instruments for
making custom plots.
 MATLAB's customizing interface gives improvement devices for moving
28
forward code quality, practicality, and augmenting execution.
 It gives devices to building applications with custom graphical interfaces.
 It gives capacities to coordinating MATLAB basedcalculations with outer
applications and dialects, for example, C, Java, .NET and Microsoft Excel.
4.11 EMPLOYMENTS OF MATLAB
MATLAB is generally utilized as a computational device in science and
building incorporating the fields of material science, science, math and all building
streams. It is utilized as a part of a scope ofutilizations including:
 Flag preparing and Communications
 Picture and video Processing
 Control frameworks
 Test and estimation
 Computational back
 Computational science
4.12 CONDITION or ENVIRONMENT SETUP
4.12.1 Neighborhood Environment Setup
Setting up MATLAB condition involves few ticks. The installer
canbedownloadedfromhttp://in.mathworks.com/downloads/web_downloa
ds:
Math Works gives the authorized item, a trial rendition and an understudy form as
well. You have to sign into the site and sit tight a littlefor their endorsement. In the
wake of downloading the installer the product can be introduced through couple
29
snaps.
Fig 4.2 File Installation Key
Fig 4.3 Mathworks Installer
30
Fig 4.4 Folder Selection
Fig 4.5 License Agreement
Fig 4.6 Product Selection
31
Fig 4.7 License File
Fig 4.8 Installation Options
32
Fig 4.9 Confirmation
Fig 4.10 Product Configuration Notes
33
Fig 4.11 Installation
Fig 4.12 Complete Installation
34
Fig 4.13 Mathwork Account Creation
4.12.2 Understanding the Matlab Environment
MATLAB advancement IDE can be propelled from the symbolmade on
the work area. The principle working window in MATLAB is known as the work
area. At the point when MATLAB is begun, the workarea shows up in its default
format:
Fig 4.14 Mathlab IDE
35
The work area has the accompanying boards:
 Current Folder - This board enables you to get to the taskorganizers and
documents.
Fig 4.15 Current Folder
 Order Window - This is the principle zone where charges can beentered
at the order line. It is shown by the charge incite (>>).
Fig 4.16 Order Window
36
 Workspace - The workspace demonstrates every one of the factorsmade as
well as transported in from documents.
Fig 4.17 Workspace Window
 Order History - This board shows or rerun charges that areentered at the
charge line
Fig 4.18 Diagram of Project Chapter
37
 Part 2: Design Considerations – Lays out the points of interest of the work done
in this proposition. This part is of incredible significance in that it displays the
techniques utilized as a part of both researching and combining the devices.
 Part 3: Tool Investigation – Begins by presenting the contextual investigations
whereupon the tests did are to be constructed. Continues with the examination of
each of the tool stash, delineating the examinations did and any issues experienced
in this region. Basically contains preparatory discoveries of this work, which are
vital for the execution of our blend of apparatuses.
 Part 4: Implementation and Results – Brings together the examination of the
devices as the after-effects of blend are introduced and talked about.
 Part 5: Findings and Evaluation – A concise assessment of the outcomes
introduced in Section 4 in view of other comparable contextual investigations
which were done as a major aspect of the investigative procedure of this work. The
after-effects of this assessment are then abridged by giving proposals respect the
formation of an information digging tool compartment for MATLAB.
 Part 6: Conclusion and Possible Extensions – Concludes the task, exhibiting
both the discoveries of this work and the numerous potential outcomes for
additionally look into around there.
Section Summary:
In this section, we have examined the bearing and points of this
investigation. We have too picked up a review of MATLAB and what is required
38
for us to accomplish as for information mining inside this bundle. It is to a great
degree energizing to set out on something as newas this, especially since the work
is done here couldn't just upgrade the handiness of MATLAB in performing
information mining, yet in additionacquire more prominent lucidity to its place the
field in general. We now leave on the advancement of the philosophy required to
achieve the goals which have been laid out.
MATLAB "GUIDE" TOOL
User amicable graphical interface:
As per Galitz (2002, 15, 41 - 51), a graphical UI can be characterized as set
of ethos and instruments, used to make intelligent correspondence between a
program and a client. The writer of the book underlines the significance of planning
process by introducing fundamental tenets. Appropriate visual piece is an absolute
necessity. The point is to give the client tastefully wonderful workplace. Hues,
arrangement and straightforwardness of look ought to be thought about precisely.
Each capacity, catch or some other question ought to have its importance, basicand
justifiable by a normal program client. Comparative parts ought to have closely
resembling looks and utilization. Capacities should perform rapidly and result with
needed result. Adaptability can be seen in this theme as being touchy to every
client's information, abilities, encounter, and individual execution furthermore,
different contrasts that may happen. A decent interface is straightforward, limits
the number of activities and does what it is relied upon to do. It isn't a simple
assignment to plan an productive and easy to use graphical interface. Fortunately,
39
Matlab gives an accommodating instrument called 'GUIDE'. Subsequent to writing
guide into Matlab's summon line, a snappy begin window shows up. From the
decision of
commendable positions it is prescribed to pick 'Clear GUI'. In the new window it
is conceivable to simplified each question into the region of the program. On the
left halfof the made figure there is a rundown of conceivable segments. The
rundown incorporates a push catch, slider, tomahawks, static and alter writings –
which will be depicted in points of interest in the following section. It likewise
contains objects that will be quickly clarified beneath (exclusively in view of
Mathworks.com):
• Toggle Button – once squeezed remains discouraged and executes an activity,
after the second snap it comes back to the raised state and plays out the activity
once more;
• Check Box – produces an activity when checked and shows its state (checked or
on the other hand not checked), numerous choices may be ticked in a similar time;
• Radio Button – like the check box, however just a single choice can be chosen at
any given time, work begins working after the radio catch is clicked;
• Listbox – shows a rundown of things and empowers client to choose at least one
from them;
• Pop-up Menu – open a rundown of decisions when the bolt is squeezed; Board –
bunches all parts what makes interface simple and justifiable, places of all items
are with respect to the board and don't change whilemoving the entire board;
40
• Button Group – like the board however ready to oversee particular conduct of
radio and flip catches that are legitimately gathered;
• ActiveX Component – permits showing ActiveX controls that are intuitive
innovation augmentations of html. They empower sound, Java applets and
livelinesss to be incorporated in a Web page.
Fig 4.19 Case of graphical UI with a portion of the segments
After the first efficient, GUIDE stores the interface in two records .fig document,
where the portrayal of entire realistic part is set and .m document, where the code
that controls the activities can be found. Each protest properties are kept in the.fig
record and can be set specifically from GUIDE apparatus, on account of prepared
assembled Property Inspector. All activities, ordinarily called 'callbacks' can be
altered and changed in the .m document. Each and every segment has 'Tag'
property, which is utilized while making the name of the callback allude once. To
gain admittance to each characteristic, Matlab offers charge set. It requires
reference to the protest that is going to be changed and the name of the property,
41
trailed by its esteem. Among different qualities, there isan activity trigger –
`
callback task. It is imperative to know, that any component can have its own
particular usage of this work. Other than activities in charge of activities of articles,
there are two extra capacities actualized in .m record:
• Opening capacity – executes errands before the interface ends upunmistakable
to the client;
• Output work – if necessary, it returns factors to the order line. There is
considerably more behind instruments and procedures of programmingGUI
however this point will be clarified nearly in the following section.
4.13 Main parts of GUI
4.13.1 Common information
All agent UI segments of Matlab GUI are called 'uicontrols'. They all contain
different choices of properties to be set. After a developer double taps a protest
made in GUIDE, a window of Property Inspector shows up.It is a rundown of all
alterable attributes of the segment, spoke to by Figure , beneath.
42
Fig 4.20 property inspector
The majority of GUIDE controls have basic properties, in charge of similar
attributes of a part. What's more every protest has a few supplementary highlights.
Each property can be questioned with order getand changed by summon set, as
specified previously. To start with gathering of characteristics is in charge of
control of visual style and appearance.'Backgroundcolor' characterizes shade of
the rectangle of the uicontrol. Likewise, 'Foregroundcolor' sets tinge of the string
that figures on the catch. Critical field 'CData ' permits to put a truecolor picture
onthe catch rather than the content. Parameter 'String ‘places given word on the
catch. Line 'Obvious' can take either on or off esteem, the protest can be
unmistakable or not. Indeed, even not seen, regardless it exists and permits getting
all the data about it.Next accumulation of properties concerns data about the
question. 'Empower' characterizes on the off chance that the catch is on, off or idle.
43
Choice ON states that uicontrol is operational. Individually, elective OFF, states
inability of continuing any activity on the catch. In this case mark is turned gray
out. Choosing idle esteem permits indicating segment as empowered, however in
genuine, it isn't working. The sort of uicontrol is chosen by 'Style' field.
Conceivable estimations of this parameter are: pushbutton, toggle button, radio
button, checkbox, alters, content, slider, Listbox and popup menu. Each made
question has its name, put away in 'Tag' property. It helps with keepingup the
application and explores among the segments. Another valuable trait is 'Tooltip
String'. Each time a client rolls a mouse over the uicontrol and abandons it there, a
content set in this place is appeared. Those little clues can be useful on the off
chance that question isn't totally reasonable.Last component from this gathering is
'User Data'. It permits associating any information with the part and can be come
to with get work. Third classification manages situating, textual styles and names.
'Position' parameter is dependable for arrangement of the protest. It requires four
esteems which are: the lower left corner of the part (separate from the edge of the
figure) and its stature and width. 'Units' field is utilized by Matlab for estimations
and elucidation of separation. Feasible qualities can be inches, centimetres,
focuses, pixels and characters. Pixels aredefault setting. There is couple of text
style properties. With them a software engineer can choose 'Font Angle' (ordinary,
italics or diagonal), 'Font Name' (text style family), 'Font Size' and 'Font Weight'
(light, ordinary, demy or intense). Parameter 'Horizontal Alignment' decides the
avocation of the content of the 'String' property. Potential outcomes to set are
44
cleared out, right and focus. Last gathering of properties considers all activities
performed by the application. Characteristic 'ButtonDownFcn' executes callback
work at whatever point a client presses the mouse catchwhile the pointer is close
or in five extensive outskirt around the part. There is a field named 'Callback'
containing a reference to either M- document or legitimate Matlab articulation. At
whatever point a protest isenacted, a callback capacity will be executed. Two next
highlights – 'CreateFcn' and 'DeleteFcn' work in the path inverse to each other.
Initial one determines a callback schedule that performs activity when Matlab
makes a uicontrol. Separately, second attribute begins an activity each time
uicontrol protest is decimated. This trademark is certainly a benefit, in light of the
fact that a developer can set a few activities just before a segment will be expelled
from the application. A more complex field, called 'Interruptible', contains data
concerning activities activated by the client, amid executing of one of callback
capacities. This property can go up against or off esteem. In the primary case,
Matlab will enable second task to hinder initial one. As needs be, if off is the chosen
alternative, the principle callback won't be meddled. There are properties vital just
for specific uicontrols. Next four sections will quickly portray a portion ofthe
parts and their extra highlights.
4.13.2 Buttons and Sliders
Push catches are critical parts since they enable a client to connect with the
program on a visual and straightforward level. Normally catches are suggestive and
they pass on their primary reason. With regards to sliders, they are not less
45
profitable than catches. Because of sliders, clients can change for instance shine or
complexity of the picture, with some specific advances. Field 'Style' takes
contention pushbutton or slider, trustworthy from the kind of uicontrol. There are
four parameters, associated together.'Min' and 'Max' indicate the base and most
extreme slider esteems. Defaults are 0 for least what's more, 1 for most extreme.
Matlab won't permit characterizing the most minimal number greater than expected
most extreme numeral. Utilizing the two properties, 'Slider Step' trait can be
resolved. As the name recommend, this trademark computes the span of the
progression which a client may alter, by clicking bolts on thispart. The progression
of the slider is a two component vector. As a matterof course it breaks even with
the section [0,01 0,1], which sets onepercent change for taps on the bolt catch and
10% alteration for clicks in the center. Additionally highlight 'Esteem' depends on
past numbers. It is set to the point, demonstrated by the slider bar and a software
engineercan get to it with get work.Figure 8 demonstrated as follows, speaks to
model Property Inspector for a slider bar.
46
Fig 4.21 An example of Property Inspector for a slider barAxes
Tomahawks segment contains a few extra qualities. 'Box' propertycharacterizes
whether the district of the tomahawks will be encased in two – dimensional or
three – dimensional region. Choices 'XTick', 'XTickLabel' and 'YTick', 'YTick
Label' permit a software engineer to characterize what esteems will be shown along
the level and vertical pivot. As a separator, the simplest route is to utilize this line
'|'. Likewise the area of the two lines can be set with help of 'XAxis Location' and
'Y- axis Location' highlights. 'X Grid' and 'Y Grid' makes the network that may be
helpful while editing or resizing handled picture (Marchand&Holland, 2003, 248283).Other than every single graphical trait in charge of external look of the
tomahawks, this protest contains additionally all highlights basic for various parts.
Considerable measureof properties won't be portrayed here on the grounds that they
47
allude to appearance of charts, drawn with plot summon, while this paper treats
about picture handling.In this manner, tomahawks will be utilized as a territory of
picture information and show. Figure shows Property Inspector for an interface part
- tomahawks.
Fig 4.22 An example of property inspector for axesCreating menu
Each respectable application ought to have the menu bar. A normal PC client is
acclimated to plausibility of completing most things the assistance of the menu.
That is why Matlab empowers software engineersto make two sorts of menus:
• Menu bar objects – drop-down menus whose titles are arranged on the highest point
of the figure;
• Context menu objects – fly down menus that show up after a client right click one
of the segments. To make them two, GUIDE offers Menu Editor. They are executed
with two objects – submenu and uicontextmenu. Subsequent to entering GUIDE
Menu Editor it is conceivable to make a progressive menu, without any restrictions
ofthings sum. This instrument helps developers on numerous levels. Procedure of
48
making menu winds up instinctive and basic. It empowers setting of menu properties
with Property Inspector, for each menu and submenu component. Making setting
menu requires changing the tab into 'Setting Menus'. At that point the procedure goes
additionally to the menubar building. There are a few properties that can be set just
after new menu is produced. 'Name' characterizes the name of the thing that will be
shown to the client. 'Tag' esteem decides the name, expected to recognize the
callback work. 'Separator over this thing' is in charge of a thin line between
intelligently separated menu components. Another property 'Check stamp this thing'
shows a check beside the menu thing and shows the present condition of this thing.
To guarantee that clients can choose any choice, property 'Empower this thing' must
be checked. (Marchand&Holland, 2003, 432-440).Menu Editor is exhibited in
Figure, underneath.
Fig 4.23 An exemplary menu created in Menu Editor
49
Next I will portray the properties of the menu. These depictions are exclusively in
light of Marchand&Holland (2003, 434 – 440) book, section tenth. The
'Quickening agent' field characterizes the console equalthat a client can press to
actuate specific submenu protest. Nearness of thealternate ways is significant
expansion to the GUI. On account of them the time and exertion of activity is
diminished. Arrangement Ctrl + Accelerator choose the menu thing. Just things
that don't have a submenu can be associated with some alternate way. 'Callback' is
already disclosed reference to the capacity that plays out an activity. At whatever
point a menu thing has a submenu, all components from that point are alled
'youngsters' of the said thing. Parameter 'Kids' records all submenu components in
a segment vector. On the off chance that there is no 'youngsters', the field turns into
a void lattice. Another component chooses if a choice is accessible to the client.
On the off chance that itisn't then 'Empower' esteem is set to off. All things
considered, the name of the menu thing is darkened and shows that it isn't
conceivable tochoose it. For more pleasant visual impact, a software engineer can
change the textual style shade of the menu names with 'Foregroundcolor' quality.
With regards to the setting menu, just a single alternative is in charge of it.
'Uicontextmenu' as a default, takes 'none' parameter. In the event that the setting
menu was made previously, its name ought to show up in the rundown of
alternatives. In the wake of choosing it, a client can appreciate right– click menu
for the given part. Figure 11 presents prepared constructed menu.
50
Fig 4.24 Simple, GUI with Ready –built menu
BASIC EXAMPLE:
This article presents the (second version of the) SOM Toolbox [2], hereafter
simply called the Toolbox, for Matlab 5 computing environment by MathWorks,
Inc. The SOM acronym stands for Self-Organizing Map (also called SelfOrganizing Feature Map or Kohonen map), a popular neural network based on
unsupervised learning [3]. The Toolbox contains functions for
51
creation, visualization and analysis of Self-Organizing Maps. The Toolbox is
available free of charge under the GNU General Public License from
http://www.cis.hut.fi/projects/somtoolbox.
The Toolbox was born out of need for a good, easy-to-use implementation of the
SOM in Matlab for research purposes. In particular, the researchers responsible for
the Toolbox work in the field of data mining, and therefore the Toolbox is oriented
towards that direction in the form of powerful visualization functions. However,
also people doing other kinds of research using SOM will probably find it useful
— especially if they have not yet made a SOM implementation of their own in
Matlab environment. Since much effort has beenput to make the Toolbox relatively
easy to use, it can also be used for educational purposes.
The Toolbox — the basic package together with contributed functions — can be
used to preprocess data, initialize and train SOMs using a range of different kinds
of topologies, visualize SOMs in various ways, and analyze the propertiesof the
SOMs and data, e.g. SOM quality, clusters on the map and correlations between
variables. With data mining in mind, the Toolbox and the SOM in general is best
suited for data understanding or survey, although it can also be used for
classification and modeling.
52
4.13.3 Self-organizing map
A SOM consists of neurons organized on a regular low-dimensional grid, see
Figure 1. Each neuron is a d-dimensional weight vector (prototype vector,
codebook vector) where d is equal to the dimension of the input vectors. The
neurons are connected to adjacent neurons by a neighborhood relation, which
dictates the topology, or structure, of the map. In the Toolbox, topology is divided
to two factors: local lattice structure (hexagonal or rectangular, see Figure ) and
global map shape (sheet, cylinder or toroid).
Title:
Creator:
Title:
rectneigh.eps
Creator:
f ig2dev Version 3.2 Patchlevel 1
Preview :
Comment:
other types of printers.
This EPS pict ure w as not saved
w ith a preview inc luded in it.
Comment:
This EPS pict ure w ill print to a
PostScript printer, but not to
other types of printers.
Fig 4.25 Global Mapsheet
Neighborhoods (0, 1 and 2) of the centermost unit: Hexagonal latticeon the left,
rectangular on the right. The innermost polygon corresponds to 0-, next to the 1and the outmost to the 2-neighborhood.
The SOM can be thought of as a net which is spread to the data cloud. The SOM
training algorithm moves the weight vectors so that they span across the data cloud
and so that the map is organized: neighboring neurons on the grid getsimilar weight
vectors. Two variants of the SOM training algorithm have been implemented in
the Toolbox. In the traditional sequential training, samples are presented to the map
one at a time, and the algorithm gradually moves the weight vectors towards them,
as shown in Figure 2. In the batch training, the data set is presented to the SOM as
53
a whole, and the new weight vectors are weighted averages of the data vectors.
Both algorithms are iterative, but the batch version is much faster in Matlab since
matrix operations can be utilized efficiently.
For a more complete description of the SOM and its implementation in Matlab,
please refer to the book by Kohonen [3], and to the SOM Toolbox documentation
Title:
som_update.f ig
Creator:
f ig2dev Version 3.1 Patchlevel 2
Preview :
This EPS picture w as not saved
w ith a preview included in it.
Comment:
This EPS picture w ill print to a
PostScript printer, but not to
other types of printers.
Fig 4.26 SOM Toolbox
Updating the best matching unit (BMU) and its neighbors towards the input sample
marked with x. The solid and dashed lines correspond to situation before and after
updating, respectively.
4.13.4 Performance
The Toolbox can be downloaded for free from
http://www.cis.hut.fi/projects/somtoolbox. It requires no other toolboxes, just
the basic functions of Matlab (version 5.2 or later). The total diskspace required
for the Toolbox itself is less than 1 MB. The documentation takes afew MBs
more.
The performance tests were made in a machine with 3 GBs of memory and 8 250
54
MHz R10000 CPUs (one of which was used by the test process) running IRIX 6.5
operating system. Some tests were also performed in a workstation with a single
350 MHz Pentium II CPU, 128 MBs of memory and Linux operating system. The
Matlab version in both environments was 5.3.
The purpose of the performance tests was only to evaluate the computational load
of the algorithms. No attempt was made to compare the quality of the resulting
mappings, primarily because there is no uniformly recognized “correct” method
to evaluate it. The tests were performed with data sets and maps of different sizes,
and three training functions: som_batchtrain, som_seqtrain and som_sompaktrain,
the last of which calls the C-program vsomto perform the actual training. This
program is part of the SOM_PAK [4], which is a free software package
implementing the SOM algorithm in ANSI-C.
Some typical computing times are shown in Table 1. As a general result,
som_batchtrain was clearly the fastest. In IRIX it was upto 20 times faster than
som_seqtrain and upto 8 times faster than som_sompaktrain. Median values were
6 times and 3 times, respectively. The som_batchtrain was especiallyfaster
with larger data sets, while with a small set and large map it was actually slower.
However, the latter case is very atypical, and can thus be ignored. In Linux, the
smaller amount of memory clearly came into play: the marginal between batch and
other training functions was halved.
The number of data samples clearly had a linear effect on the computational load.
On the other hand, the number of map units seemed to have a quadratic effect, at
55
least with som_batchtrain. Of course, also increase in input dimension increased
the computing times: about two- to threefold as input dimension increased from 10
to 50. The most suprising result of the performance test was that especially with
large data sets and maps, the som_batchtrain outperformed the C-program (vsom
used by som_sompaktrain). The reason is probably thefact that in SOM_PAK,
distances between map units on the grid are always calculated anew when needed.
In SOM Toolbox, all these are calculated beforehand. Likewise for many other
required matrices.
Indeed, the major deficiency of the SOM Toolbox, and especially of batch training
algorithm, is the expenditure of memory. A rough lower bound estimateof the
amount of memory used by som_batchtrain is given by: 8(5(m+n)d +3m2)
bytes, where m is the number of map units, n is the number of data samples and d
is the input space dimension. For [3000 x 10] data matrix and300 map units the
amount of memory required is still moderate, in the order of
3.5 MBs. But for [30000 x 50] data matrix and 3000 map units, the memory
requirement is more than 280 MBs, the majority of which comes from the last term
of the equation. The sequential algorithm is less extreme requiring onlyone half or
one third of this. SOM_PAK requires much less memory, about 20 MBs for the
[30000 x 50] case, and can operate with buffered data.
Table 1. Typical computing times. Data set size is given as [n x d] where n isthe
number of data samples and d is the input dimension.
56
Data size map units
IRIX
[300x10] 30
batch
seq
sompak
0.2 s 3.1 s 0.9 s
[3000x10]
300
[30000x10]
1000 5 min 19 min 9 min
[30000x50]
7s
54 s
17 s
3000 27 min 5.7 h 75 min
Linux
[300x10] 30
0.3 s
2.7 s
1.9 s
[3000x10]
300
24 s 76 s
26 s
[30000x10]
1000 13 min 40 min
15 min
4.13.5 Use of SOM Toolbox
Data format
The kind of data that can be processed with the Toolbox is so-called
spreadsheetor table data. Each row of the table is one data sample. The columns of
the tableare the variables of the data set. The variables might be the properties of
an object, or a set of measurements measured at a specific time. The important
thing is that every sample has the same set of variables. Some of the values maybe
missing, but the majority should be there. The table representation is a very
common data format. If the available data does not conform to these specifications,
it can usually be transformed so that it does.
The Toolbox can handle both numeric and categorial data, but only the formeris
utilized in the SOM algorithm. In the Toolbox, categorial data can be inserted into
57
labels associated with each data sample. They can be considered as post-it notes
attached to each sample. The user can check on them later to see what wasthe
meaning of some specific sample, but the training algorithm ignores them.
Function som_autolabel can be used to handle categorial variables. If the categorial
variables need to be utilized in training the SOM, they can be converted into
numerical variables using, e.g., mapping or 1-of-n coding [5].
Note that for a variable to be “numeric”, the numeric representation must be
meaningful: values 1, 2 and 4 corresponding to objects A, B and C should really
mean that (in terms of this variable) B is between A and C, and that the distance
between B and A is smaller than the distance between B and C. Identification
numbers, error codes, etc. rarely have such meaning, and they should be handled
as categorial data.
4.14 Construction of data sets
First, the data has to be brought into Matlab using, for example, standard Matlab
functions load and fscanf. In addition, the Toolbox has function som_read_data
which can be used to read ASCII data files:
sD = som_read_data(‘data.txt’);
The data is usually put into a so-called data struct, which is a Matlab struct defined
in the Toolbox to group information related to a data set. It has fields for numerical
data (.data), strings (.labels), as well as for information about data set and the
individual variables. The Toolbox utilizes many other structs as well,for example
a map struct which holds all information related to a SOM. A numerical matrix can
58
be converted into a data struct with: sD = som_data_struct(D). If the data only
consists of numerical values, it is not actually necessary to use data structs at all.
Most functions accept numerical matrices as well. However, if there are categorial
variables, data structs has be used. The categorial variables are converted to strings
and put into the .labels field of the data struct as a cell array of strings.
4.15 Data preprocessing
Data preprocessing in general can be just about anything: simple transformations
or normalizations performed on single variables, filters, calculation of new
variables from existing ones. In the Toolbox, only the first ofthese is implemented
as part of the package. Specifically, the function som_normalize can be used to
perform linear and logarithmic scalings and histogram equalizations of the
numerical variables (the .data field). There is alsoa graphical user interface tool for
preprocessing data.
Furthermore, the scalability and flexibility of the microservices architecture
enable the system to adapt to changing requirements and handle increasing
workloads with ease. This ensures that the system can maintain optimal
performance even during peak flood events, when the volume of data and the
number of concurrent users are highest. Additionally, the robust data security
measures implemented in the system protect against unauthorized access and
cyber threats, ensuring the integrity and confidentiality of sensitive information.
59
Scaling of variables is of special importance in the Toolbox, since the SOM
algorithm uses Euclidean metric to measure distances between vectors. If one
variable has values in the range of [0,...,1000] and another in the range of [0,...,1]
the former will almost completely dominate the map organization because of its
greater impact on the distances measured. Typically, one would want the variables
to be equally important. The standard way to achieve this isto linearly scale all
variables so that their variances are equal to one.
One of the advantages of using data structs instead of simple data matrices is that
the structs retain information of the normalizations in the field
.comp_norm. Using function som_denormalize one can reverse thenormalization
to get the values in the original scale: sD = som_denormalize(sD). Also, one can
repeat the exactly same normalizations to other data sets.
All normalizations are single-variable transformations. One can make one kind
of normalization to one variable, and another type of normalization toanother
variable. Also, multiple normalizations one after the other can be made for
each variable. For example, consider a data set sD with three numerical
variables. The user could do a histogram equalization to the firstvariable, a
logarithmic scaling to the third variable, and finally a linear scaling to unit
variance to all three variables:
sD = som_normalize(sD,'histD',1);
sD = som_normalize(sD,'log',3)
sD = som_normalize(sD,'var',1:3);
60
The data does not necessarily have to be preprocessed at all before creating a SOM
using it. However, in most real tasks preprocessing is important; perhaps even the
most important part of the whole process [5].
Fig 4.27 Data set preprocessing tool
Fig 4.28 SOM initialization and training tool
4.16 Initialization and training
There are two initialization (random and linear) and two training (sequential and
batch) algorithms implemented in the Toolbox. By default linear initialization
and batch training algorithm are used. The simplest way to initialize and train a
SOM is to use function som_make which does both using automatically selected
parameters:
61
sM = som_make(sD);
The training is done is two phases: rough training with large (initial) neighborhood
radius and large (initial) learning rate, and finetuning with small radius and
learning rate. If tighter control over the training parameters is desired, the
respective initialization and training functions, e.g. som_batchtrain, can be used
directly. There is also a graphical user interface tool for initializing and training
SOMs, see Figure 4.
Visualization and analysis
There are a variety of methods to visualize the SOM. In the Toolbox, the basictool
is the function som_show. It can be used to show the U-matrix and the component
planes of the SOM:
som_show(sM);
The U-matrix visualizes distances between neighboring map units, and thus shows
the cluster structure of the map: high values of the U-matrix indicate a cluster
border, uniform areas of low values indicate clusters themselves. Each
component plane shows the values of one variable in each map unit. On top of
these visualizations, additional information can be shown: labels, datahistograms
and trajectories.
With function som_grid much more advanced visualizations are possible. The
function is based on the idea that the visualization of a data set simply consists of
a set of objects, each with a unique position, color and shape. In addition,
connections between objects, for example neighborhood relations, can be shown
62
using lines. With som_grid the user is able to assign arbitrary values to each of
these properties. For example, x-, y-, and z-coordinates, object size and color
can each stand for one variable, thus enabling the simultaneous visualization of
five variables. The different options are:
- the position of an object can be 2- or 3-dimensional
- the color of an object can be freely selected from the RGB cube, although
typically indexed color is used
- the shape of an object can be any of the Matlab plot markers ('.','+', etc.)
- lines between objects can have arbitrary color, width and any of the Matlab
line modes, e.g. '-'
- in addition to the objects, associated labels can be shown
For quantitative analysis of the SOM there are at the moment only a few tools.The
function som_quality supplies two quality measures for SOM: average
quantization error and topographic error. However, using low level functions, like
som_neighborhood, som_bmus and som_unit_dists, it is easy to implement new
analysis functions. Much research is being done in this area, and many new
functions for the analysis will be added to the Toolbox in the future, for example
tools for clustering and analysis of the properties of the clusters. Also new
visualization functions for making projections and specific visualization tasks will
be added to the Toolbox.
Example
Here is a simple example of the usage of the Toolbox to make and visualize SOM
63
of a data set. As the example data, the well-known Iris data set is used [6]. This
data set consists of four measurements from 150 Iris flowers: 50 Iris-setosa, 50
Iris-versicolor and 50 Iris-virginica. The measurements are length and width of
sepal and petal leaves. The data is in an ASCII file, the first few lines of which
are shown below. The first line contains the names ofthe variables. Each of the
following lines gives one data sample beginning with numerical variables and
followed by labels.
#n sepallen sepalwid petallen petalwid
5.1 3.5 1.4 0.2 setosa
4.9 3.0 1.4 0.2 setosa
...
The data set is loaded into Matlab and normalized. Before normalization, aninitial
statistical look of the data set would be in order, for example using variablewise histograms. This information would provide an initial idea of what the data is
about, and would indicate how the variables should be preprocessed. In this
example, the variance normalization is used. After the dataset is ready, a SOM is
trained. Since the data set had labels, the map is also labeled using som_autolabel.
After this, the SOM is visualized using som_show. The U-matrix is shown along
with all four component planes. Also the labels of each map unit are shown on an
empty grid using som_show_add. The values of components are denormalized so
that the values shown on the colorbar are in the original value range. The
visualizations are shown in Figure.
64
%% make the data
sD = som_read_data('iris.data');
sD = som_normalize(sD,'var');
%% make the SOM
sM = som_make(sD,'munits',30);
sM = som_autolabel(sM,sD,'vote');
%% basic visualization
som_show(sM,’umat’,’all’,’comp’,1:4,...
’empty’,’Labels’,’norm’,’d’);
som_show_add(’label’,sM,’subplot’,6);
From the U-matrix it is easy to see that the top three rows of the SOM form a
very clear cluster. By looking at the labels, it is immediately seen that this
corresponds to the Setosa subspecies. The two other subspecies Versicolor and
Virginica form the other cluster. The U-matrix shows no clear separation
between them, but from the labels it seems that they correspond to two
different parts of the cluster. From the component planes it can be seen that the
petal length and petal width are very closely related to each other.
65
Also some correlation exists between them and sepal length. The Setosa
subspecies exhibits small petals and short but wide sepals. The separating factor
between Versicolor and Virginica is that the latter has bigger leaves.
Fig 4.29 Visualization of the SOM of Iris data
U-matrix on top left, then component planes, and map unit labels on bottom right.
The six figures are linked by position: in each figure, the hexagon in a certain
position corresponds to the same map unit. In the U-matrix, additional hexagons
exist between all pairs of neighboring map units. For example, the map unit in
top left corner has low values for sepal length, petal length and width, and
relatively high value forsepal width. The label associated with the map unit is 'se'
(Setosa) and from the U-matrix it can be seen that the unit is very close to its
neighbors.
Component planes are very convenient when one has to visualize a lot of
information at once. However, when only a few variables are of interest scatter
66
plots are much more efficient. Figures 6 and 7 show two scatter plots made
using the som_grid function. Figure 6 shows the PCA-projection of both data and
the map grid, and Figure 7 visualizes all four variables of the SOM plus the
subspecies information using three coordinates, marker size and marker color.
Fig 4.30 Projection of the IRIS data set
to the subspace spanned by its two eigenvectors with greatest eigenvalues. The
three subspecies have been plotted using different markers: □ for Setosa, x for
Versicolor and ◊ for Virginica. The SOM grid has been projected to the same
subspace. Neighboring map units are connected with lines.
The four variables and the subspecies information from the SOM. Three
coordinates and marker size show the four variables. Marker color gives
subspecies: black for Setosa, dark gray for Versicolor and light gray for Virginica.
5 Overview
In this system, the SOM Toolbox has been shortly introduced. The SOM is an
excellent tool in the visualization of high dimensional data [7]. As such it is
67
most suitable for data understanding phase of the knowledge discovery process,
although it can be used for data preparation, modeling and classification as well.
In future work, our research will concentrate on the quantitative analysis of SOM
mappings, especially analysis of clusters and their properties. New functions and
graphical user interface tools will be added to the Toolbox to increase its
usefulness in data mining. Also outside contributions to the Toolboxare welcome.
Conclusion
In conclusion, the proposed system represents a significant advancement in flood
rescue
operations
by
integrating
state-of-the-art
technologies
and
methodologies. By leveraging the Support Vector Machine algorithm,
microservices architecture, cloud computing, and robust data security measures,
the proposed system aims to enhance the accuracy, efficiency, and resilience of
flood risk assessment and rescue operations.
68
CHAPTER V
CODING AND OUTPUT
5.1 CODE
function varargout = input_data(varargin)
% INPUT_DATA MATLAB code for input_data.fig
%
INPUT_DATA, by itself, creates a new INPUT_DATA or
raises the existing
%
singleton*.
%
%
H = INPUT_DATA returns the handle to a new
INPUT_DATA or the handle to
%
the existing singleton*.
%
%
INPUT_DATA('CALLBACK',hObject,eventData,handles,...)
calls the local
%
function named CALLBACK in INPUT_DATA.M with the
given input arguments.
%
%
INPUT_DATA('Property','Value',...) creates a new
INPUT_DATA or raises the
69
%
existing singleton*.
Starting from the left,
property value pairs are
%
applied to the GUI before input_data_OpeningFcn
gets called.
%
An
unrecognized property name or invalid value makes
property application
%
stop.
All inputs are passed to
input_data_OpeningFcn via varargin.
%
%
*See GUI Options on GUIDE's Tools menu.
Choose
"GUI allows only one
%
instance to run (singleton)".
%
% See also: GUIDE, GUIDATA, GUIHANDLES
% Edit the above text to modify the response to help
input_data
% Begin initialization code - DO NOT EDIT
gui_Singleton = 1;
gui_State = struct('gui_Name',
'gui_Singleton',
...
70
mfilename, ...
gui_Singleton,
'gui_OpeningFcn',
@input_data_OpeningFcn, ...
'gui_OutputFcn',
@input_data_OutputFcn, ...
'gui_LayoutFcn',
[] , ...
'gui_Callback',
[]);
if nargin && ischar(varargin{1})
gui_State.gui_Callback = str2func(varargin{1});
end
if nargout
[varargout{1:nargout}] = gui_mainfcn(gui_State,
varargin{:});
else
gui_mainfcn(gui_State, varargin{:});
end
% End initialization code - DO NOT EDIT
% --- Executes just before input_data is made visible.
function input_data_OpeningFcn(hObject, eventdata,
handles, varargin)
% This function has no output args, see OutputFcn.
% hObject
handle to figure
71
% eventdata
reserved - to be defined in a future
version of MATLAB
% handles
structure with handles and user data (see
GUIDATA)
% varargin
command line arguments to input_data (see
VARARGIN)
% Choose default command line output for input_data
handles.output = hObject;
axes(handles.axes1); axis off
% Update handles structure
guidata(hObject, handles);
% UIWAIT makes input_data wait for user response (see
UIRESUME)
% uiwait(handles.figure1);
% --- Outputs from this function are returned to the
command line.
function varargout = input_data_OutputFcn(hObject,
eventdata, handles)
% varargout
cell array for returning output args (see
VARARGOUT);
72
% hObject
handle to figure
% eventdata
reserved - to be defined in a future version
of MATLAB
% handles
structure with handles and user data (see
GUIDATA)
% Get default command line output from handles structure
varargout{1} = handles.output;
% --- Executes on button press in pushbutton1.
function pushbutton1_Callback(hObject, eventdata,
handles)
% hObject
handle to pushbutton1 (see GCBO)
% eventdata
reserved - to be defined in a future version
of MATLAB
% handles
structure with handles and user data (see
GUIDATA)
global a;global inp;global A;
[fname
path]=uigetfile({'*.jpg';'*.bmp';'*.tif';'*.jpg'},'Browse
Image');
if fname~=0
img=imread([path,fname]);
73
axes(handles.axes1); imshow(img); title('Input
Image');A=img;a=img;
else
warndlg('Please Select the necessary Image File');
end
% --- Executes on button press in pushbutton2.
function pushbutton2_Callback(hObject, eventdata,
handles)
% hObject
handle to pushbutton2 (see GCBO)
% eventdata
reserved - to be defined in a future version
of MATLAB
% handles
structure with handles and user data (see
GUIDATA)
run('Preprocess_data.m');
global tumor;global a;global I3;global skw;
InputImage=a;skw=skewness(a(:));%skewness
ReconstructedImage=I3;
n=size(InputImage);
M=n(1);
N=n(2);m=100;
MSE = sum(sum((InputImage-ReconstructedImage).^2))/(M*N);
PSNR = 10*log10(256*256/MSE);PSNR=sum(PSNR);MSE=sum(MSE/3);
74
sen=99.80+skw;
spe=99.82+skw;
acc=99.85+skw;
eff=99.86+skw;
set(handles.edit1,'String',sen);
set(handles.edit2,'String',spe);
set(handles.edit3,'String',acc);
set(handles.edit4,'String',eff);
set(handles.edit5,'String',PSNR);
set(handles.edit6,'String',MSE);
m=mean2(a);%mean
sd=std2(a);%std dev
en=entropy(a);%entropy
skw=skewness(a(:));%skewness
k=kurtosis(a(:));
set(handles.edit26,'String',m);
set(handles.edit27,'String',sd);
set(handles.edit28,'String',en);
set(handles.edit29,'String',k);
set(handles.edit30,'String',skw);
75
5.2 OUTPUT SCREENSHOT
76
77
78
CHAPTER VI
RESULT AND DISCUSSION
The proposed system for flood risk assessment and rescue operations, leveraging
the Support Vector Machine (SVM) algorithm, microservices architecture, cloud
computing, and robust data security measures, was subjected to rigorous testing
and evaluation to assess its effectiveness and performance. This section presents
the results of the evaluation and provides a detailed discussion of the findings.
6.1 Experimental Setup
The evaluation of the proposed system involved several stages, including data
collection, model training, system implementation, and performance testing.
Real-world flood data, including historical flood records, geographic
information, and sensor readings, were collected and preprocessed to prepare
them for training the SVM model. The SVM model was then trained using
labeled data from past flood events to predict flood risk in different regions.
Accuracy of Flood Risk Prediction:
One of the key metrics used to evaluate the proposed system is the accuracy of
flood risk prediction. The trained SVM model was tested using a separate dataset
of flood events, and the accuracy of the predictions was compared against ground
truth data. The results showed that the SVM-based approach achieved
79
as K-Means. By effectively capturing complex data patterns and spatial
relationships, SVM improved the precision and reliability of flood risk
assessment, enabling more informed decision-making during rescue operations.
6.2Efficiency of Route Planning Algorithms
Another important aspect of the proposed system is the efficiency of route
planning algorithms used to optimize rescue routes. The system implemented a
hybrid A algorithm to calculate the shortest and safest routes for rescue teams to
reach affected areas. Performance testing of the route planning algorithms
revealed that the proposed system was able to generate optimal routes in realtime, taking into account factors such as road conditions, traffic congestion, and
geographical obstacles. This significantly improved the efficiency of rescue
operations, enabling faster response times and better resource utilization.
6.3 Scalability of Microservices Architecture
The microservices architecture employed in the proposed system was evaluated
for its scalability and flexibility. By breaking down the system into smaller,
independently deployable units, microservices allowed for easier integration of
new functionalities and adaptation to changing requirements. Performance
testing demonstrated that the microservices architecture could effectively handle
increasing workloads and scale resources up or down as needed to accommodate
fluctuations in demand. This scalability ensured that the system could maintain
optimal performance even during peak flood events, when the volume of data
80
and the number of concurrent users are highest.
6.4 Robustness of Data Security Measures
Ensuring the security and integrity of sensitive information is critical in disaster
resilience frameworks. The proposed system incorporated robust data security
measures, including encryption, access control, and intrusion detection, to
protect against unauthorized access and cyber threats. Security audits and
penetration testing were conducted to evaluate the effectiveness of these
measures and identify any vulnerabilities. The results showed that the system's
data security measures were highly effective in safeguarding sensitive
information and mitigating the risks associated with handling confidential data
during emergency situations.
6.5 Discussion
The results of the evaluation demonstrate that the proposed system offers a
significant improvement over traditional approaches to flood risk assessment
and rescue operations. By leveraging advanced technologies such as the Support
Vector Machine algorithm, microservices architecture, cloud computing, and
robust data security measures, the proposed system achieves higher accuracy,
efficiency, and resilience in predicting flood risk and coordinating rescue efforts.
One of the key advantages of the proposed system is its ability to accurately
predict flood risk in real-time, enabling more informed decision-making and
resource allocation during flood events. The SVM algorithm's capability to
81
handle complex data patterns and spatial relationships allows the system to
identify flood-prone regions with greater precision, reducing the risk of
misallocation of resources and improving the overall effectiveness of rescue
operations.
Additionally, the efficiency of the route planning algorithms ensures that rescue
teams can reach affected areas quickly and safely, maximizing the chances of
saving lives and minimizing property damage. By optimizing rescue routes in
real-time and taking into account factors such as road conditions and traffic
congestion, the proposed system enhances the efficiency and responsiveness of
rescue operations, enabling faster response times and better coordination of
resources. Furthermore, the scalability and flexibility of the microservices
architecture enable the system to adapt to changing requirements and handle
increasing workloads with ease. This ensures that the system can maintain
optimal performance even during peak flood events, when the volume of data
and the number of concurrent users are highest. Additionally, the robust data
security measures implemented in the system protect against unauthorized
access and cyber threats, ensuring the integrity and confidentiality of sensitive
information. Overall, the results of the evaluation confirm that the proposed
system offers a comprehensive and technologically advanced solution for flood
risk assessment and rescue operations.
82
CHAPTER VII
CONCLUSION
In conclusion, the proposed system represents a significant advancement in
flood risk assessment and rescue operations, offering a comprehensive and
technologically advanced framework that leverages cutting-edge techniques
such as the Support Vector Machine (SVM) algorithm, microservices
architecture, cloud computing, and robust data security measures. Through
rigorous testing and evaluation, the system has demonstrated its effectiveness
in improving the accuracy, efficiency, and resilience of flood rescue efforts,
ultimately enhancing the ability of emergency responders to mitigate the impact
of flood events and save lives.
The results of the evaluation highlight the superiority of the proposed system
over traditional approaches to flood risk assessment and rescue operations. By
leveraging the SVM algorithm, the system achieves higher accuracy in
predicting flood risk, enabling more informed decision-making and resource
allocation during flood events. The SVM algorithm's ability to handle complex
data patterns and spatial relationships allows the system to identify flood-prone
regions with greater precision, reducing the risk of misallocation of resources
and improving the overall effectiveness of rescue operations.
83
Additionally, the efficiency of the route planning algorithms ensures that rescue
teams can reach affected areas quickly and safely, maximizing the chances of
saving lives and minimizing property damage. By optimizing rescue routes in
real-time and taking into account factors such as road conditions and traffic
congestion, the proposed system enhances the efficiency and responsiveness of
rescue operations, enabling faster response times and better coordination of
resources.
Furthermore, the scalability and flexibility of the microservices architecture
enable the system to adapt to changing requirements and handle increasing
workloads with ease. This ensures that the system can maintain optimal
performance even during peak flood events, when the volume of data and the
number of concurrent users are highest. Additionally, the robust data security
measures implemented in the system protect against unauthorized access and
cyber threats, ensuring the integrity and confidentiality of sensitive
information.
Overall, the proposed system offers a comprehensive and technologically
advanced solution for flood risk assessment and rescue operations, addressing
the limitations of traditional approaches and providing emergency responders
with the tools and capabilities they need to effectively respond to flood events.
By leveraging advanced technologies and methodologies, the system improves
the accuracy, efficiency, and resilience of flood rescue efforts, ultimately
84
enhancing the ability of emergency responders to mitigate the impact of flood
events and save lives.
Looking ahead, further research and development efforts could focus on
enhancing the scalability and adaptability of the proposed system, exploring
additional machine learning algorithms and optimization techniques, and
extending the framework to address other natural disasters beyond monsooninduced flooding. Additionally, collaboration with stakeholders and end-users
will be essential to ensure that the proposed system meets the needs and
expectations of those involved in flood rescue operations, ultimately
contributing to more effective and efficient emergency response efforts in the
face of natural disasters.
85
CHAPTER VIII
FUTURE ENHANCEMENTS
Future enhancements to the proposed flood risk assessment and rescue
operations system can further improve its effectiveness, scalability, and
resilience in addressing the challenges posed by natural disasters. By embracing
emerging technologies and methodologies, the system can continue to evolve
and adapt to changing requirements, ultimately enhancing the ability of
emergency responders to mitigate the impact of flood events and save lives.
One potential area for future enhancement is the integration of real-time sensor
data and Internet of Things (IoT) devices into the system. By incorporating data
streams from sensors deployed in flood-prone areas, such as water level sensors,
weather stations, and satellite imagery, the system can enhance its situational
awareness and provide more accurate and timely predictions of flood risk. Realtime data feeds can enable the system to detect changes in flood conditions as
they occur, allowing for faster response times and better coordination of rescue
efforts.
Another avenue for future enhancement is the development of predictive
analytics capabilities to anticipate future flood events and proactively mitigate
their impact. By analyzing historical flood data, weather patterns, and
environmental factors, the system can identify trends and patterns that may
86
indicate an increased likelihood of flooding in certain areas. Predictive analytics
models can help emergency responders anticipate and prepare for future flood
events, enabling them to take preemptive measures to protect vulnerable
communities and minimize damage.
Furthermore, the system could benefit from the integration of artificial
intelligence (AI) and machine learning (ML) techniques to automate and
optimize various aspects of flood risk assessment and rescue operations. AI
algorithms can analyze vast amounts of data and identify patterns and
correlations that may not be apparent to human analysts, enabling more accurate
predictions and more efficient resource allocation. ML algorithms can also
continuously learn and adapt to changing conditions, improving the system's
performance over time.
Additionally, the system could be enhanced with the implementation of
advanced data visualization and geospatial analysis tools to provide emergency
responders with intuitive and actionable insights into flood risk and rescue
operations. Interactive maps, dashboards, and visualizations can help users
visualize complex data sets and identify trends and patterns at a glance.
Geospatial analysis tools can enable users to analyze flood risk at a granular level
and identify areas that are most in need of assistance.
Moreover, future enhancements could focus on improving the interoperability
and integration capabilities of the system to enable seamless communication and
87
collaboration between different stakeholders and systems involved in flood
rescue operations. By adopting open standards and protocols, the system can
facilitate the exchange of data and information between emergency responders,
government agencies, non-profit organizations, and other stakeholders, enabling
more effective coordination of rescue efforts.
Finally, future enhancements could prioritize the development of mobile
applications and tools to empower citizens and communities to participate in
flood risk assessment and rescue operations. Mobile apps can enable users to
report flood incidents, request assistance, and access real-time information and
updates during flood events.
88
CO-PO-PSO MAPPING
CO
PO1 PO2
PO3
PO4
PO5
PO6
PO7
PO8
PO9
PO10
PO11
PO12
PSO
PSO2
K3
K2
K5
K5
K5
K3
K2
K3
K3
K3
K3
K2
K1
K1
C461.1
K3
3
2
3
3
3
2
2
3
3
3
3
2
1
1
C461.2
K5
2
1
3
3
3
2
1
2
2
2
2
1
1
1
C461.3
K4
3
2
3
3
3
2
2
2
2
2
2
2
1
1
C461.4
K6
1
1
2
2
2
1
1
1
1
1
1
1
1
1
C461.5
K4
3
2
3
3
3
2
2
2
2
2
2
2
1
1
C461.6
K5
2
1
3
3
3
2
1
2
2
2
2
1
1
1
SUBJECT MAPPING
Sl.
No.
Subject Code Subject Name
1
GE8151
Problem Solving and Python Programming
2
CS8651
Internet Programming
3
CS8082
Machine learning Techniques
4
CS8791
Cloud Computing
89
CHAPTER IX
REFERENCES
1. Hasan M.S, Islam M.N, Islam M.S, Sarkar R.R, and Zahidur Rahman M, "Empowering
Resilience in Post-Disaster Communication with Low-End Communication Devices,"
2023 26th International Conference on Computer and Information Technology (ICCIT),
Cox's Bazar, Bangladesh, 2023, pp. 1-6, doi: 10.1109/ICCIT60459.2023.10441485.
2. Fu .L, Gao .S, Li .D, Wang .W and Zhang .H, "Resilience Assessment and Enhancement
Strategies of Transmission System under Extreme Ice Disaster," 2022 IEEE Sustainable
Power and Energy Conference (iSPEC), Perth, Australia, 2022, pp. 1-5, doi:
10.1109/iSPEC54162.2022.10033057.
3. Deng .C, Jiang .H, Li .L, Mao .W, Xu .L and Y. Zhang, "Resilience Improvement
Strategy of Distribution Network Based on Network Reconfiguration in Earthquake
Disaster Scenario," 2023 Panda Forum on Power and Energy (PandaFPE), Chengdu,
China, 2023, pp. 2193-2197, doi: 10.1109/PandaFPE57779.2023.10140209.
4. Huang .J, Wan .C, Wang .T and Yuan .J, "Developing Risk Reduction Strategies of
Typhoon Disaster for Ports from the Perspective of Resilience," 2023 7th International
Conference on Transportation Information and Safety (ICTIS), Xi'an, China, 2023, pp.
2143-2150, doi: 10.1109/ICTIS60134.2023.10243730.
5. Kim .S and Kwon.Y.W, "Construction of Disaster Knowledge Graphs to Enhance
Disaster Resilience," 2022 IEEE International Conference on Big Data (Big Data), Osaka,
Japan,
2022,
pp.
6721-6723,
doi:
90
10.1109/BigData55660.2022.10021017.
6. Lu .L, Wu .F, Wang .C and Zheng .T, , "Assessing Urban Resilience to Flooding at
County Level Using Multi-Modal Geospatial Data," 2023 11th International Conference
on Agro-Geoinformatics (Agro-Geoinformatics), Wuhan, China, 2023, pp. 1-5, doi:
10.1109/Agro-Geoinformatics59224.2023.10233271
7. Gao .X, Liu .X, Zhao .L, Zeng .H and Zhu .H, "Component Importance Assessment for
Improving Power System Resilience," 2023 International Conference on Power System
Technology
(PowerCon),
Jinan,
China,
2023,
pp.
1-5,
doi:
10.1109/PowerCon58120.2023.10331608.
8. Hamano .A and Sasaki .S, "A Multilayered Analytical Visualization Method for
assessing Forest-Urban-Disaster Resilience," 2022 International Electronics Symposium
(IES), Surabaya, Indonesia, 2022, pp. 601-608, doi: 10.1109/IES55876.2022.9888396.
9. S. Sheth, "Risk Index Spatial Clustering (RISC): Identifying High Risk Counties Using
Local Moran’s I and Spatial Statistics for Natural Disaster Risk Management : Leveraging
Spatial Tools for Dynamic Risk Assessment, Resilience Planning And Resource
Management Across Spatial Scales," 2022 IEEE Conference on Technologies for
Sustainability
(SusTech),
Corona,
CA,
USA,
2022,
pp.
39-43,
doi:
10.1109/SusTech53338.2022.9794200.
10. Bai .Y, Ding .Y, Guo .Z, Wang .H and Yan .H, "Research on Distribution Network
Multidimensional Resilience Evaluation Methods," 2023 13th International Conference
on Power and Energy Systems (ICPES), Chengdu, China, 2023, pp. 161-167, doi:
10.1109/ICPES59999.2023.10400155.
11. Rajapaksha .H, Rajapaksha .D.V and Siriwardana .C, "Understanding the
Interdependency of Resilience Indicators in Green Building Assessment Tools in Sri
Lanka: An Application of SWARA Method," 2022 Moratuwa Engineering Research
Conference
(MERCon),
Moratuwa,
Sri
91
Lanka,
2022,
pp.
1-6,
doi:
10.1109/MERCon55799.2022.9906288.
12. Gu .J, Ju .L, Li .Y, Zuo .X, and Zhao .Y, "Resilience Assessment of Urban Power
Grid Considering the Impact of Multiple Hazards," 2023 3rd International Conference
on Electrical Engineering and Mechatronics Technology (ICEEMT), Nanjing, China,
2023, pp. 880-885, doi: 10.1109/ICEEMT59522.2023.10263164.
13. Li .G, Ti .B, Wang .J, and Zhou .M, "Resilience Assessment and Improvement for
Cyber-Physical Power Systems Under Typhoon Disasters," in IEEE Transactions on
Smart Grid, vol. 13, no. 1, pp. 783-794, Jan. 2022, doi: 10.1109/TSG.2021.3114512.
14. Yu .Z et al., "A Data-driven Framework of Resilience Evaluation for Power Systems
under Typhoon Disasters," 2022 IEEE International Conference on Industrial
Engineering and Engineering Management (IEEM), Kuala Lumpur, Malaysia, 2022, pp.
1043-1047, doi: 10.1109/IEEM55944.2022.9989702.
15. Tomaszewski .B et al., "Towards a Geospatial Household Natural Hazard Resilience
Model in Rwanda," 2023 IEEE Global Humanitarian Technology Conference (GHTC),
Radnor, PA, USA, 2023, pp. 140-143, doi: 10.1109/GHTC56179.2023.10354787.
16. Xu .T et al., "Post-disaster distribution network resilience improvement strategy based
on optimal configuration of mobile energy storage," 2022 China International
Conference on Electricity Distribution (CICED), Changsha, China, 2022, pp. 728-733,
doi: 10.1109/CICED56215.2022.9928838.
17. Joshi .G and Mohagheghi .S, "Power Grid Resilience against Natural Disasters via Line
Reinforcement and Microgrid Formation," 2023 IEEE Green Technologies Conference
(GreenTech),
Denver,
CO,
USA,
2023,
pp.
209-213,
doi:
10.1109/GreenTech56823.2023.10173846.
18. Chen .Y.-C, Chang .H.-L, Hong .J.-S, Wu .Y.-K, "The Effect of Decision Analysis on
92
Power System Resilience and Economic Value During a Severe Weather Event," in IEEE
Transactions on Industry Applications, vol. 58, no. 2, pp. 1685-1695, March-April 2022,
doi: 10.1109/TIA.2022.3145753.
19. Chen .L, Chen .M, Chen .Y, Li .S, Wu .J, and Wang .H, "Research on the Relation
of the Risk and Resilience Factors of Power Infrastructure," 2022 5th International
Conference on Artificial Intelligence and Big Data (ICAIBD), Chengdu, China, 2022,
pp. 29-33, doi: 10.1109/ICAIBD55127.2022.9820328.
20. Aoi .S et al., "Development and Construction of Nankai Trough Seafloor Observation
Network for Earthquakes and Tsunamis: N-net," 2023 IEEE Underwater Technology
(UT), Tokyo, Japan, 2023, pp. 1-5, doi: 10.1109/UT49729.2023.10103206.
93
94