International Journal of Advancements in Research & Technology, Volume 2, Issue 7, July-2013 ISSN 2278-7763 258 Fault detection of Induction Motor using Envelope Analysis* RincyRaphael, Bipin PR 1 Applied Electronics, ICET, Ernakulam, India; 2Electronics, ICET, Ernakulam, India. Email:rincyraphel@gmail.com. ABSTRACT This paper presents a new approach to diagnose broken rotor bar fault in induction motor using envelope analysis. A dynamic model of induction motor is developed using MATLAB/SIMULINK. Wavelet Transform and envelope analysis used to obtain statistical parameters maximum, median, minimum. Depending on these parameters, Support Vector Machine predicts the machine is healthy or not. Keywords: Broken Rotor Bar, Envelope Analysis, Discrete Wavelet Transform, Support Vector Machine. 1 INTRODUCTION I NDUCTION motor is a most commonlyused industrial load and guzzles a key part of overall electrical consumption. Induction motors, especially the asynchronous motors, play a vital part in the field of electromechanical energy conversion.The main advantages of induction motorsare costeffectiveness, reliability and higher efficiency. Electric induction motors have the ability to be connected directly to the AC source. This can be an important cost savernotonly in household uses but also in industrial uses. An induction motor is easy to program for its numerous uses. The initial cost of installing will be high, but it will save money in the long term because of the low maintenance cost and durability of the product. Moreover, it is a flexible design which allows innovations and newer technologies to be incorporated easily without incurring extra costs. Different fault occurs in induction motorbecause of environmental stress and many other reasons. Induction motors are commonly controlled by contactors, which are electromagnetic switches that are highly sensitive to voltage depressions and momentary service interruptions. Voltage depressions are huge problems for many industries, and it is probably the major pressing power quality problems today. The Voltage depressions caused by faults on the system affect the performance of induction motors, in terms of the production of both transient currents and transient torques [1]. It is essential to minimize the effect of the voltage dip on both the induction motor and more significantly on the process where the motor is used. Large torque peaks may cause damage to the shaft or equipment connected to the shaft. Some common reason for voltage depressions are lightning strikes in power lines, equipment failures, accidental contact power lines, and electrical machine starts. Despite being a short duration between 10 milliseconds to 1 second event during which are duction in the RMS voltage magnitude takes place, a small reduction in the system voltage can cause serious consequences .This unexpected interruption at the industrial plant disturbs the cost, product quality and safety. Thus any fault affecting induction motor has a drastic effect. Hence fault detection and diagnosis have much importance. It is highly essential to protect the induction motor against failures. It is better to detect faults before motor crashes completely. Hugh Douglas, Pragasen Pillay, andAlireza K. Ziarani discussed about a new algorithm which is introduced to do motor current signature analysis of induction machines operating during transients [3].The algorithm is able to extract the amplitude, phase andfrequency of a single sinusoid embedded in a nonstationary waveform. The algorithm is applied to the detection of broken rotor bars in induction machines during startup transients. The fundamental component of current, which varies in amplitude, phase, and frequency, is extracted using the algorithm. The residual current is then analyzed using wavelets for the detection of broken rotor bars. This method of condition monitoring does not require parameters such as speed or number of rotor bars, is not load dependent and can be applied to motors that operate continuously in the transient mode. BurakOzpineci, Leon M Tolbert they proposed a modular Simulink implementation of an induction machine model is described in a step-by-step approach [8]. With the modular system, each block solves one of the model equations. All of the machine parameters are accessible for control and verification purposes. After the implementation, examples are given with the model used in different drive applications, such as open-loop constant V/Hz control and indirect vector control. S.K. Ahamed, ArghyaSarkar, M.Mitra and S. Senguptaexplains that errors may creep in when broken rotor bar fault of induction motor is detected using conventional FFT based approach as small sideband harmonics are completely masked by the power frequency [10]. This paper unveils a new wavelet transform and envelope analysis based broken rotor bar fault detection technique, free from above mentioned drawback. Wavelet transform has been performed onthe windowed steady state current signal at no-load to obtain detailed and approximate coefficients at different spectral bands. Then the envelopes of the reconstructed signal corresponding to spectral bands below 50Hz were determined using Hilbert transform. It has been observed that faulty motor produces higher statistical parameters- mean, standard deviation, median of the envelopes than those for the healthy one. IJOART Copyright © 2013 SciResPub. IJOART International Journal of Advancements in Research & Technology, Volume 2, Issue 7, July-2013 ISSN 2278-7763 KhalafSalloumGaeid and Hew Wooi Ping proposed a method which explains that the wavelet is considered as powerful tools in the fault detection and diagnosis of induction motors [9]. Many wavelet classes can be generated by different kinds of mother wavelets and can be constructed by filters banks. The improvement of fault detection and diagnosis can be exploiting the wavelet properties to get high detection and diagnostics effectiveness. Theories of wavelet need to be pushed forward to insure best choosing of mother wavelet. The wavelet index can distinguish correctly between the faults and healthy induction motor. Matlab/Simulink package for both simulations and practical experiments in the diagnostic of induction machines with wavelet. A variety of faults can occur within three phase induction motor during a course of normal operation. These faults can lead to a potentially catastrophic failure if undetected. The common internal faults can be mainly categorized in to two groups [7]. Mechanical faults: • Broken rotor bars • Bearing damage • Dynamic eccentricity Electrical Faults • Single or multiple phase short circuits. • Open phase. • Abnormal connections. In this paper broken rotor bar or cracked rotor end ring are analyzed with the help of signal processing techniques. The first step to detect the broken rotor bar fault is to develop analysis techniques that can be used to diagnose the observed current signal to get useful information [4].Data acquisition is carried out by running the simulation. The data acquisition phase is followed by a sequence of steps consisting of signal processing, envelope analysis and classification process. The statistical parameters of the rotor current signal in frequency domain are obtained and the parameters which can represent the fault features well are extracted. Then classification of fault is done with the help of Support Vector Machine. 259 the stationary frame to the synchronously rotating frame (dq) using (2). In place of voltages there may be currents or flux linkages.Where ‘ϒ’ is the transformation angle. (1) (2) 2. 2phase to 3-phase (dq-abc) conversion: This conversion does the opposite of the abc-dq conversion for the current variables using (3) and (4) respectively by following the same implementation techniques as before. (3) IJOART (4) The dynamic equations in arbitrary reference frame which is rotating at speed ω in the direction ofrotor rotation. 1. When ω is equal to zero, the reference frame is fixed in the stator (qs-ds ref. frame). 2. When ω is equal to ωe, the reference frame is fixed on the synchronously rotating referenceframe. 3. When ω is equal to ωr, the reference frame is fixed in the rotor. That is, the reference frame isrotating at speed of ω r . The dynamic equations used are: Electrical system equations: (5) 2 MODELING OF INDUCTION MOTOR Induction motor is modeled using mat lab / simulink program. This model is described in space vector formulation in synchronously rotating arbitrary d-q reference frame associated with the frequency of the stator excitation. This model is obtained considering the induction motor as 6 magnetically coupled circuits [1]. Under balanced conditions, the sum of stator currents as well as rotor currents is zero. The induction motors are modeled in d-q variables. The d-q model [6] uses two windings for each stator and rotor of the induction motor. Only four independent variables are needed to model the induction motor. A transformation of variables can be used. The power invariant two-axes transformation is used and is defined as: 1.3-phase to 2-phase (abc-dq) conversion: To convert 3-phase voltages to voltages in the 2-phase synchronously rotating frame, they are first converted to 2-phase stationary frame (α β) or in the movement with an arbitrary speed using (1) and then from Copyright © 2013 SciResPub. (6) (7) Flux linkage-current relations: (7a) (7b) IJOART International Journal of Advancements in Research & Technology, Volume 2, Issue 7, July-2013 ISSN 2278-7763 Where Mechanical system equations: (8) 3 SIMULINK MODEL OF INDUCTION MOTOR 260 3.1 Model for Broken Bar Fault of Induction Motors When the induction motor parameter changes there occur interior faults. This paper provides one type of interior fault model [5]. With the help of the quantitative relation among machine parameters, fault types and fault extent, it is possible to indirectly reflect interior faults by setting parameter changes of induction motor model. Considering the greatly possible broken bar failure, a mathematical model is obtained. Suppose that Z 2 is the number of rotor bars and n is the number of broken bars. A cage rotor with n broken bars can be analyzed as a three phase wound rotor with its impedance getting increased by ∆Z (∆Z = (∆R/s + j∆X)) dependent on the broken bar number. This means that the difference in the mutual linkage inductance coefficients between different windings can be neglected [2]. The transformation value of rotor resistance is R’ 2 =R’ B +R’ R The transformation value of rotor resistance after failure is (9) IJOART (10) Fig.1 Block diagram of induction motor model in the arbitrary frame The model of the induction motor in d-q variables is represented by (5), (6), (7) and (8). Currents or flux linkages can be used as state variables. In this paper the flux linkages are chosen as state variables and it was verified in the Matlab/Simulink software environment. The block diagram of a three-phase induction motor supplied by a three-phase power supply is shown in Fig. 1. The variable ω = ω s represents the speed of the common reference frame used. This speed is an arbitrary speed. It consists of five major blocks: the 3-phasevoltages, abc-dq conversion, dq-abc conversion, induction machine d-q model (Im_dq_model) and the mechanical system (mech sys) blocks. Fig.3. Faulty Model 4 PROPOSED TECHNIQUE Fig.2. Simulink Model of Induction Motor Copyright © 2013 SciResPub. Rotor current is taken from the simulink model of induction motor. Then Discrete Wavelet Transform is performed on the current signal with frequency below 50Hz using db6 mother wavelet. The statistical parameters are obtained by envelop analysis. Then the envelopes of detailed coefficients are obtained, it is observed that the faulty motor produces higher statistical parameters maximum, median and minimum than those for the healthy motor. Then the statistical parameters are classified with the help of support vector machine.Model of healthy and faulty induction motor are taken IJOART International Journal of Advancements in Research & Technology, Volume 2, Issue 7, July-2013 ISSN 2278-7763 as reference for training the Support Vector Machine.Envelope analysis not only enhances the fault parameters, remaining power frequency leakage if any is also removed. Wavelets are functions that can be used to decompose signals, similar to how to use complex sinusoids in a Fourier Transform to decompose signals. The wavelet transform computes the inner products of analyzed signals and a family of wavelets. Wavelets are more suitable for analyzing nonstationary or transient signals, than time domain or frequency domain analysis. DWT is well-suited for multi-resolution analysis. The DWT decomposes high frequency components of a signal with fine time resolution but coarse frequency resolution and decomposes low frequency components with fine frequency resolution but coarse time resolution .DWT based multi-resolution analysis help better understand a signal and is useful in feature extraction application such as fault detection. Multi-resolution analysis also can help in removing unwanted components in the signal, such as noise. The family of Daubechies wavelets was chosen as the basis functions for the decomposition. 261 TABLE II CLASSIFICATION RESULTS USING SVM TABLEIII 4 RESULTS Using d and q variables in a synchronously rotating reference frame, a generalized dynamic model of an induction motor is developed. A mathematical model of the fault is obtained and is incorporated with the d-q model. A Support Vector Machine fault prediction is trained with the help of parameters obtained from healthy and faulty simulink models. IJOART TABLE I CLASSIFICATION RESULTS USING NN Copyright © 2013 SciResPub. DWT-ENVELOPE ANALYSIS RESULTS Comparison between ANN classification and SVM classi- IJOART International Journal of Advancements in Research & Technology, Volume 2, Issue 7, July-2013 ISSN 2278-7763 262 fication is carried out and is shown in table I and table II. In both cases healthy and faulty models are considered. The table I shows ANN classification is good for low value of rotor resistance and give wrong output for higher values. But SVM classification is good for higher values of rotor resistance and gives wrong output for lower values. In this paper provide accurate result for both higher and lower rotor resistances shown in table III. 5 CONCLUSION In this present paper dynamic model of induction motor has been developed using d and q variables in a synchronously rotating frame. The broken rotor bar fault is detected using wavelet transform and envelope analysis. Support Vector Machine is used to determine whether induction motor is healthy or not. REFERENCES [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] B. ArunaKumari, K. Naga Sujatha, K. Vaisakh, Dept. of Electrical Engineering, “Assessment Of Stresses On Induction Motor Under Different Fault Conditions”, Journal of Theoretical and Applied Information Technology, Vol.50, 2009. Y.amuna. K. Moorthy, Pournami S. Chandran, Rishidas S. Asst. Professor, Dept of Electronics & Communication College of Engineering, Trivandrum “Motor Current Signature Analysis by Multi-resolution Methods using Support Vector Machine”, 978-1-4244-9477-4/11/$26.00 ©2011 IEEE 096 H. Douglas, P. Pillay, A. K. Ziarani, “A new algorithm for transient motor current signature analysis using wavelets”University of Cape Town Department of Electrical Engineering Rondebosch 7701 Cape Town, South Africa. B. Ayhan, M. Y. Chow, and M. H. Song, “Multiple signature processingbased fault detection schemes for broken rotor bar in induction motors”, IEEE Trans. Energy Convers., vol. 20, no. 2, pp. 336343, Jun. 2005. S.S.Mulukutla, E.M.Gulachenski, “A critical survey of considerations in maintaining process continuity during voltage dips while protecting motors with reclosing and bustransferpractices,”IEEE, 1991. J. Lee, P. Pillay And R. G. Harley, “D,Q Reference Frames for the Simulation of Induction Motors”, Department of Electrical Engineering, University of Natal,Electric Power Systems Research, Regular Papers, 8 (1984/85). Cusido, L. Romeral, A. Garcia, J.A. Rosero , J.A. Ortega, “Fault detection in Induction Machines by using Continuous and Discrete Wavelet Decomposition” MCIA research group published in Power Electronics and Applications, 2007 European Conference on 2-5 Sept. 2007. BurakOzpineci, M. Leon, Tolbert, “Simulink Implementation of Induction Machine Model- A Modular Approach”, Department of Electrical and Computer Engineering the University of Tennessee IEEE Transactions on Power Electronics 2003. KhalafSalloumGaeid and Hew Wooi Ping, “Wavelet Fault Diagnosis of Induction Motor” University of Malaya MalaysiaMATLAB for Engineers – Applications in Control, Electrical Engineering, ITand Robotics. S.K. Ahamed1,* Arghya Sarkar2, M.Mitra3 and S. Sengupta4 “Broken Rotor Bar Fault Detection of Induction Motor through Wavelet Transform and Envelope Analysis” Published in Computing, Electronics and Electrical Technologies (ICCEET), 2012 International Conference IJOART Copyright © 2013 SciResPub. IJOART