Sci.Int.(Lahore),26(4),1447-1456,2014 ISSN 1013-5316; CODEN: SINTE 8 1447 CLASSIFICATION ANALYSIS OF POWER SYSTEM TRANSIENT DISTURBANCES WITH SOFTWARE CONCEPTS Aslam P. Memon1, M. Aslam Uqaili3, Zubair A. Memon3, Waqar Adil2, M. Usman Keerio4 1 Mehran University of Engineering, & Technology, Jamshoro & Department of Electrical, Quaid-e- Awam U.E.S.T, Nawabshah, Sindh, Pakistan; aslam@quest.edu.pk 2,4 Department of Electrical Engineering, Quaid-e-Awam University of Engineering, Science & Technology, Nawabshah, Sindh, Pakistan. 3 Department of Electrical Engineering, Mehran University of Engineering, & Technology, Jamshoro, Sindh, Pakistan .ABSTRACT: Power system transients cause serious disturbances in the reliability, safety and economy of the power system network. These disturbances occur in a few cycles, which are difficult to be identified and classified by digital measuring and recording instrumentations. The transient signals are the transitory for which the frequencies as well as varying time information are compulsory for the analytic purposes. Various types of transients indicate various behaviors and measuring characteristics, but it is vital, first to detect and classify the type of fault and then to mitigate them. In digital signal processing, Fast Fourier transformation is a powerful technique traditionally utilized to measure the disturbances of such signals, but it is more suitable for periodic analysis where time information of the signal is not necessary. In this work power system transient disturbances are detected, analyzed and classified with the help of time-frequency analysis technique of discrete wavelet transformation with multiresolution analysis (MRA) algorithm choosing the most suitable wavelet function as mother wavelet. This proposed methodology possesses the ability to de-noise and decompose the various types of transient using Matlab/Simulink and Wavelet toolbox and the simulation results prove their simplicity, accurateness and effectiveness for the detection and classification of power system transient signal disturbances. Keywords: Power Quality, Power System Transient, Discrete Wavelet Transform, Symlet Mother Wavelet, Development of Numerical, Simulink and GUI models .INTRODUCTION The transients occur due to switching, lighting strikes, and various types of faults and other intended or unintended causes. They become the detrimental reasons of power system components which suffer from the enormous amount of currents and voltages. Transients are also known as voltage disturbances less than sag or swell and caused by sudden variations in electrical power system [1-3]. PST is a sub-category of electrical power quality (EPQ), arising as a result of variation in sinusoidal nature of power system (PS) supply voltage which is said to be nonrepetitive, random and causes equipment failure. Various types of PSTs indicate various behaviors and measuring characteristics. The severe magnitude and energy content make them necessary to be investigated for their detection and classification to facilitate the various power system protection equipment [1-4]. Therefore, it is imperative to understand the characteristics of the transient to identify its causes and sources in order to develop appropriate mitigation methodology to prevent the damage of sensitive equipment [5]. To analyze transient signals, we can either generate them by simulation methods or collect them in a real time environment. Real time collection of transient data is expensive and time consuming. The next step in the analysis is to choose an effective signal processing technique which is normally a mapping the signals in he frequency domain and is known as Fast Fourier transformation (FFT). FFT is unfortunately not effective in detecting the aperiodic elements of the signals. A time-frequency mapping known as wavelets transform (WT) is preferred over other techniques in such time varying cases [6-8]. 1.1 Literature Review Fourier transform, which is considered as conventional signal analysis methods, and uses sine waves with infinite support, find limited usage in the analysis of transient disturbances. In general, transients are non-stationary and short-term waveforms that change the sinusoidal voltage and current waveforms in an aperiodic manner. More advanced techniques, such as the wavelet transform, are more suitable. In these techniques waves with compact (finite) support are used [8-10] and these techniques reveal signal properties while retaining its time information. This is known as fractal analysis [11-12]. As a result of analyzing transient with these techniques, one can obtain features of transients that can be used in a classifier for differentiating their various types. PST is the important part of power quality issue which is aperiodic in nature. The chances of occurrence PST disturbances are uncertain and hence they are difficult to be measured, recorded and analyzed. Due to unavailability of real time transient data, it is again difficult for researchers and academicians to focus on research related to transient disturbance phenomenon. FFT deals only in frequencydomain periodic signals hence its signal analysis technique is not capable to detect localized and classify the PST disturbances [13-15]. 1448 ISSN 1013-5316; CODEN: SINTE 8 Therefore, it is imperative to understand the various characteristics of PST disturbances to identify its causes and sources in order to develop appropriate mitigation methodology to prevent the damage of sensitive equipment. This work proposes the detection and classification of PST signal disturbances with the help of time-frequency technique of discrete wavelet transformation (DWT). The multiresolution analysis (MRA) as the algorithm is suggested and the suitable mother wavelet is also chosen from some of the mostly used mother wavelet functions which produce the feature extraction data for the classification of PST signals. All the major categories of PSTs defined by IEEE-1159 are developed using the mathematical equations and graphical user interfacing (GUI) in Matlab. 1. POWER QUALITY (PQ) AND POWER SYSTEM TRANSIENT DISTURBANCE (PSTD) The IEEE Standards Coordinating Committee 22 (IEEE SCC22) has led the main effort to coordinate power quality standards. It has classified power quality problems as [1-3]: Disturbances Waveform distortion Voltage unbalance Voltage flicker Power frequency variations Transients are sub-category of power disturbances which have a high frequency. The IEEE 1159 standard defines two types of transients [1-3]; Impulsive transients Oscillatory transients Fig. 01: Voltage transients An electrical transient is the outward manifestation of a sudden change in circuit conditions, as when a switch opens or closes or a fault occurs on a system. Transient period is very short relative to the time spent in steady state operations, but is extremely important because greatest stress occurs on circuit components due to excessive currents and voltages which cause damage to the circuit. Transients can be classified depending upon their wave shape, their origin, and their mode of generation [1-3]. Wave shape based classification includes: Oscillatory (Ringing wave) Impulsive Origin based classification includes: Atmospheric origin i.e. lightning Switching origin i.e. switching operation and faults. On the other hand mode of generation based classification includes: Sci.Int.(Lahore),26(4),1447-1456,2014 Electromagnetic transients. (Interaction between capacitive and inductive elements) Electromechanical transients Interaction between capacitive and inductive elements results in travelling waves and oscillations [16]. This work mainly deals with the electromagnetic transients. 2. DIGITAL SIGNAL PROCESSING (DSP) AND WAVELET TRANSFORM (WT) Although root mean square (RMS) method is not an inherent signal processing technique, yet it is the most used tool. RMS gives a good approximation of the fundamental frequency, amplitude profile of a waveform. A great advantage of this algorithm is its simplicity, speed of calculation and less requirement of memory, because rms can be stored periodically instead of per sample [17]. However, its dependency of window length is considered as a disadvantage. One cycle window length will give better results in terms of profile than a half cycle window. Moreover, rms does not distinguish fundamental frequency, harmonics or noise components. On the other hand, rms voltage profiles are used for event analysis and automatic classification as proposed in [18]. Monitoring of power quality problem using advanced DSP techniques comprises: Processing of stationary signals Processing of non-stationary signals Processing techniques of power quality data are further classified based on their method of analyzing the waveform data. These are “Parametric or model based methods for processing of PQ data”, and “Non-parametric or transformation based processing of PQ waveform data”. Parametric or model based methods for processing of PQ data include: Harmonic or sinusoidal model Multiple Signal Classification model ESPRIT method Kalman Filtering based method Non-parametric or transformation based processing of PQ waveform data include: Fourier Transform Short Time Fourier Transform Wavelet Transform In this work transformation based processing techniques have been discussed and utilized. Traditionally, the Fourier transforms permits mapping signals from time domain to the frequency domain by decomposing the signals into several frequency components. This technique is criticized in that the time information of transients is totally lost, although the accuracy of frequency components is high. Fourier transform does not fit the analysis of transients owing to the non-stationary property of its signals in both time and frequency domains. Wavelet transform generally offers this facility [19] 3.1 Wavelet analysis The wavelet analysis calculates the correlation between the signal under consideration and a wavelet function ψ (t). The Sci.Int.(Lahore),26(4),1447-1456,2014 ISSN 1013-5316; CODEN: SINTE 8 similarity between the signal and the analyzing wavelet function is computed separately for different time intervals, resulting in a two dimensional representation. The analyzing wavelet function ψ (t) is also referred to as the mother wavelet. An analyzing function ψ (t) is classified as a wavelet if the following mathematical criteria are satisfied [20]: 1. A wavelet must have finite energy E (t ) 2 (1) The energy E equals the integrated squared magnitude of the analyzing function ψ (t) and must be less than infinity. 2. If Ψ(f) is the Fourier transform of the wavelet ψ(t), the following condition must hold C ( f ) f 0 2 df (2) This condition implies that the wavelet has no zero frequency component (Ψ (0) = 0), i.e. the mean of the wavelet ψ (t) must equal zero. This condition is known as the admissibility constant. The value of Cψ depends on the chosen wavelet. 3. For complex wavelets the Fourier transform Ψ (f) must be both real and vanish for negative frequencies. For computer implementations discrete wavelet transform (DWT) is utilized as: W m, n a0 k a 0 m nb0 xk m a0 k 1 m (3) a0 , b a0 nb0 , m and n are the integer numbers provided a 0 1 and b0 0 [19]. m Where a m Due to this process redundancy of continuous form must be eliminated hence a 0 and b0 be selected as to from orthogonal basis by satisfying the condition as a0 2 and b0 1 . This requirement invites us to use multiresolution analysis (MRA), which is also known as multiresolution wavelet method (MWM). In this method original signal xt is decomposed into different scales resolutions and the mother wavelet function ψt 2 d n 2t n is To formulate a numerical model for transient signal’s waveform generation, the resulting transient voltage is considered to be consisted of two components, normal power system voltage, and additive transient voltage. Therefore, following relation is implemented to generate power system transient event waveform [6-7]. VResultingwaveform VPower system VTransient (4) 4.1.1.1 Oscillatory Transient dt 1449 chosen with V sin(2ft ) u.a. sin(2f osct1 ).e ( decay factor).t1 (5) The second term in Eq. (4.2) models the transient oscillations having frequency fosc. The transient element is superimposed on the supply voltage at any required instant of time with the help of unit step function u as defined in Eq. (6). 1 : t if t - t 1 0 u 0 : t if t - t 1 0 4.1.1.2 IMPULSIVE TRANSIENT V sin(2f t ) [u.a.(e t1 e t1 )] Table 1: IEEE standards for power system transient signals S. No. Categories 1 Transients 1.1 n t 2 c n 2t n known as scaling function, where n d n and c n are squared sum able sequences [21-22]. 3. METHODOLOGY 4.1 Transient Waveform Generation MATLAB software is used to simulate the power system transient disturbances numerically. 4.1.1 Numerical models (7) 4.1.2 Standard adopted for transient signal generation All the simulations comply with the IEEE 1159 standard guidelines in terms of magnitude and spectral contents like rise, fall and duration. Table 1 provides information regarding typical spectral content, duration, and magnitude where appropriate for each category of electromagnetic phenomena [1-3]. 4.1.3 Matlab implementation of numerical models A user friendly GUI is designed in the Layout editor and programmed in M-file editor of MATLAB. The GUI is shown in Fig 2. By this GUI user can first simulate the transient signals with an option of variety of spectral contents. function (6) Impulsive 1.1.1 Nanosecond 1.1.2 Microsecond 1.1.3 Millisecond 1.2.1 Oscillatory Low frequency Medium 1.2 1.2.2 1.2.3 frequency High frequency Typical spectral content Typical duration 5-ns rise 1- s rise 0.1-ms rise <50 ns 50 ns–1 ms >1 ms --- <5 kHz 5–500 kHz 0.5– 0.3–50 ms 20μs 0–4 pu 5μs 0–4 pu 5MHz Typical voltage magnitude ----- 0–8 pu 1450 ISSN 1013-5316; CODEN: SINTE 8 Sci.Int.(Lahore),26(4),1447-1456,2014 3.3 CLASSIFICATION OF PSTD BY FEATURE EXTRACTION To extract the features from various coefficients of wavelet transform of the signals at different levels, the total number of decomposition levels is the first thing to know. The number of levels is found by Eq. (8) J log 2 N Fig. 02: Graphical User Interface (GUI) The basic functions of the GUI are: Generation of transient signals Saving the generated signals on hard drive Performing time frequency analysis by initializing the wavelet toolbox 4.2 SUITABLE WAVELET FUNCTION FOR DETECTION OF PSTD Four wavelet functions are found to be frequently used in the area of power system transients, namely; db4, coif4, bior3.1 and sym5. These mostly used wavelet functions are assessed on the basis of perfect reconstruction algorithm, for their suitability for accurate detection of PSTD. As shown in Fig. (3), in this algorithm the signal after reconstruction is compared with the signal before decomposition in order to investigate the suitability of wavelet function to be used. (8) Where the signals to be decomposed has ‘N’ samples. The following features are calculated at different decomposition levels to classify the transients. Percentage Energy at each decomposition level. RMS value of detail components of wavelet coefficient at maximum decomposition level. 4.3.1 Energy feature vector The energy of the signal can be related to the energy in each of the wavelet resolution levels by ‘Parseval’s theorem’, as shown in Eq. (9) 2 2 2 j 0 k (9) The energy carried by a signal at some decomposition level is found by calculating the norm of the wavelet coefficient obtained at that level. The feature vector of energy at a decomposition level is formed as shown below in Eq. (10) ESignal [ cAJ 2 2 cDJ 2 2 cD1 2 2 ........... cDJ 1 (10) 2 T 2 ] Where cAJ 1/ 2 2 [ cA J ] 2 [ cD J ] 2 K - cD J Fig. 03: Decomposition and reconstruction by filters of wavelet transform f (t ) dt k a(k ) d j (k ) 1/ 2 2 K - 4.3.2 RMS feature vector Root mean square value of voltage can be calculated as: T Vrms 1 2 v dt T 0 E Signal T (11) Since the energy at some decomposition level is nothing but the norm of wavelet coefficient at that level, the Eq. (10) can be again utilized to find the RMS values of individual wavelet coefficients. Fig. 04: Steps involved in selection of suitable wavelet function Sci.Int.(Lahore),26(4),1447-1456,2014 ISSN 1013-5316; CODEN: SINTE 8 1451 Fig. 5: Some selected waveforms from simulations 5 RESULTS 5.1 SIMULATION RESULTS A total of 32 signals have been simulated for various categories of power system transient, sixteen of which are impulsive and remaining are oscillatory. 5.2 SUITABLE WAVELET FUNCTION The Table 2 shows points that are observed related to performance of each wavelet function. 5.3 RESULTS OF DETECTION OF PSTD BY MRA Multiresolution analysis of simulated PSTD signals reveals sub-band frequency coefficients along with the time information of the disturbances also. These coefficients are outputs of high pass filters of wavelet transform, give an indication of PSTD. As shown in fig 6, a high frequency element is successfully extracted from the signal ‘s’ at first level ‘d 1’ but its localization in time is not good, i.e. the end point of disturbance is not prominent. Furthermore, MRA gives a wide range of coefficient values, and at the time of transient event peak coefficient value is 1452 Fig. 6 ISSN 1013-5316; CODEN: SINTE 8 : Low frequency oscillatory transient and five levels of decomposition by sym5 observed at a particular level, which is sufficient to detect the PSTD, as in fig 6 it is observed at third level ‘d3’. Similarly, at higher decomposition levels, the value of detail coefficient becomes unusable due to very poor localization of high frequency elements, i.e. high frequency elements starts spreading over entire time axis, as in fig 6 it is observed in 5th level ‘d5’. Table 3 presents the required Sci.Int.(Lahore),26(4),1447-1456,2014 sampling frequency to detect different categories of PSTD with the help of MRA of DWT. All categories of PSTD with different sampling rates are simulated and hence when analyzed by MRA algorithm, they produce wavelet transform sub-band of different frequency ranges. The required sampling rate for detection is found by Nyquist theorem. 5.4 CLASSIFICATION BY EXTRACED FEATURES The RMS and percent energy values of each PSTD are plotted, considering RMS and percent energy values in yaxis and number of decomposition levels in x-axis. 5.4.1 Low frequency oscillatory transient The simulated PSTD of this kind ranges from 500 Hz to 4250 Hz. The RMS and energy feature vectors are extracted and plotted across the 11 levels as shown in fig 7, 8 There is a noticeable trend common to all low frequency transient signals from 5 th level to 11th level. For low frequency transient signal we get the highest RMS and percentage energy values at 7th level. 5.4.2 Medium frequency oscillatory transient The simulated PSTD of this kind ranges from 50000 Hz to 450000 Hz. The RMS and energy feature vectors are extracted and plotted across the 14 levels as shown in fig 9, 10. There is a noticeable trend common to all medium frequency transient signals from 11th to 14th level. For medium frequency transient signal we get the highest RMS and percent energy values at 14th level. Table 2: Reconstruction performance of different wavelet functions Impulsive Wavelet Function Minimum error Db4 2.22×10 Bior3.1 0 Coif4 2.22×10 Sym5 0 -16 -16 Oscillatory Better reconstruction results Minimum error Medium/Fast 3.33×10 Fast 0 Medium/Fast 8.88×10 Slow/Medium/Fast 0 -16 Better reconstruction results Medium/Fast Medium/Fast -16 Medium/Fast Slow/Medium/ Fast Table 3: Detection results by MRA of various categories of PSTD Transient Type Oscillatory Transient Impulsive transient Highest value coefficient Frequency range Sampling frequency for detection Low d3 0.62kHz – 1.25kHz 2.5kHz Medium d3 62.5kHz – 0.125MHz 0.25MHz High d1 1MHz-2MHz 4MHz Milliseconds d2 1.25KHz-2.5KHz 5kHz Microseconds d3 (0.31 – 0.62)MHz 1.24MHz Nanoseconds d3 15.62MHz – 31.2 MHz 62.5MHz Sci.Int.(Lahore),26(4),1447-1456,2014 Fig 7: ISSN 1013-5316; CODEN: SINTE 8 RMS Pattern for signal with low frequency oscillatory transients 1453 Fig. 10: Energy Pattern for signal with medium frequency oscillatory transients 5.4.3 High frequency oscillatory transient The simulated PSTD of this kind ranges from 500000 Hz to 2750000 Hz. The RMS and energy feature vectors are extracted and plotted across the 16 levels as shown in fig 11, 12. There is a noticeable trend common to all high frequency transient signals from 13th to 16th level. For high frequency transient signals we get the highest RMS and percent energy values at 16th level. Fig 8: Energy Pattern for signal with low frequency oscillatory transients Fig. 11: RMS Pattern for signal with high frequency oscillatory transients Fig. 9: RMS Pattern for signal with medium frequency oscillatory transients Fig. 12: Energy Pattern for signal with high frequency oscillatory transients 1454 ISSN 1013-5316; CODEN: SINTE 8 Sci.Int.(Lahore),26(4),1447-1456,2014 5.4.4 Millisecond impulsive transient The simulated PSTD of this kind ranges from rise time of 0.1 msec and duration of 6 msec to a rise time of 1 msec with duration of 2 msec. The RMS and energy feature vectors are extracted and plotted across the 11 levels as shown in fig 13, 14. There is a noticeable trend common to all millisecond impulsive transient signals from 5th to 11th levels. For millisecond transient signals we get the highest RMS and percent energy values at 11th level and 7th level respectively. Fig. 15: RMS Pattern for signal with microsecond impulsive transients Fig. 13: RMS Pattern for signal with milliseconds impulsive transients Fig. 16: Energy Pattern for signal with microsecond impulsive transients vectors are extracted and plotted across the 14 levels as shown in fig 17, 18. There is a noticeable trend common to all microsecond impulsive transient signals from 6 th to 16th level. For microsecond transient signals we get the highest RMS and percent energy values at 15th level. Fig. 14: Energy Pattern for signal with milliseconds impulsive transients 5.4.5 Microsecond impulsive transient The simulated PSTD of this kind ranges from rise time of 100 µsec and duration of 700 µsec to a rise time of 1 µsec with duration of 50 µsec. The RMS and energy feature vectors are extracted and plotted across the 14 levels as shown in fig 15, 16. There is a noticeable trend common to all microsecond impulsive transient signals from 7th to 14th level. For microsecond transient signals we get the highest RMS and percent energy values at 14th level and 11th level respectively. 5.4.6 Nanosecond impulsive transient The simulated PSTD of this kind ranges from rise time of 100 µsec and duration of 700 µsec to a rise time of 1 µsec with duration of 50 µsec. The RMS and energy feature Fig. 17: RMS Pattern for signal with Nanoseconds impulsive transients Sci.Int.(Lahore),26(4),1447-1456,2014 ISSN 1013-5316; CODEN: SINTE 8 Fig. 18: Energy Pattern for signal with nanoseconds impulsive transients 6. CONCLUSIONS This research work presents investigation of wavelet analysis to detect and classify power system transients. The investigation relies on MRA algorithm of DWT, in which transient signals are decomposed into various timefrequency resolution levels. First of all the PSTD have been simulated using numerical models, complying with all the spectral constraints given by IEEE-1159 standards. Before applying MRA, the question about selecting a particular wavelet function (mother wavelet) for the wavelet analysis has been taken into consideration. A suitable wavelet function is then used in MRA of DWT, in which the transient disturbance in the power system voltage is detected at a certain resolution level in the detail component of DWT coefficient, where as approximate component of DWT coefficient shows low frequency power system voltage. This research work reveals the fact that for identifying the type of transient disturbance the values of a certain extracted feature at a particular resolution level is not sufficient to distinguish one kind of disturbance from the other. Numerical models assist the validation of power transient events characterization tool in the proposed method, and can be utilized to contribute to the development of power transient waveform simulation. Proposed MATLAB based GUI of Power transient simulator is flexible because it can be used to generate and save the PSTD data. The following points are noted in the simulation of PSTD: By using step-function, the occurrence of transient disturbances can be manipulated over the entire length of the power system voltage. Numerical model for oscillatory transient consists of single exponential for controlling the decay time of ringing/oscillations , while that of impulsive transient contains double exponential; one for manipulating the rise of impulse, and second for deciding the decaying duration of the impulse. For fast transient of duration in microseconds and nanoseconds, increasing the simulation period increases the number of samples (data points) and hence increases the computational complexity and time of wavelet analysis then after. Symlet 5 (Sym5) is found to be more adequate as a mother wavelet for detection of power system transients. A perfect 1455 reconstruction criterion of filters is adopted to compare db4, coif, bior3.1, and sym5 wavelet functions, because wavelet transformation is considered as passing the given signal from a set of conjugate quadrature mirror high pass and low pass filters. Sym5 gives least error between RMS values of signal before decomposition and after reconstruction. The following points are noted in this investigation: Db4 is efficient in medium and high frequency disturbances. Bior3.1 is efficient in high frequency disturbances. Coiflet is efficient in medium and high frequency disturbances. Sym5 is efficient in low medium and high frequency disturbances. For the accurate detection of PST, the frequency bands of each decomposed coefficient of DWT are crucial. Keeping in view, the magnitude and localization of disturbance captured at a certain detail coefficient of DWT, it is concluded that: Waveforms of level 1 to 3 detail coefficients can be analyzed for detection of PSTD. To detect oscillatory transient of low frequency to high frequency range, a sampling frequency of 2.5 kHz – 4 MHz is required respectively. To detect impulsive transient of milliseconds duration to nanoseconds duration range, a sampling frequency of 5 kHz – 62.5 MHz is required respectively. For classification of PSTD, energy and RMS values of signals are calculated at various resolution levels and following conclusive points are noted: For identifying the type of transient disturbance the values of a certain extracted feature at a particular resolution level is not sufficient to distinguish one kind of disturbance from the other. From energy and RMS profile it is evident that for classification of PSTD, the level 5 to 16 can be used. REFERENCES [1] M. H. J. Bollen, IEEE Industry Applications Society, IEEE Power Electronics Society, and IEEE Power Engineering Society, “Understanding Power Quality Problems”: Voltage Sags and Interruptions. New York: IEEE Press, 2000. [2] Dugan Roger. et al, “Electrical Power Systems Quality”, Second Edition, McGraw-Hill Publication, 2002. [3] IEEE Recommended Practice For Monitoring Electric Power Quality, IEEE Std. 159-1995. [4] G.T. Heydt, and A. W. Galli, (1997) “Transient power quality problems analyzed using wavelets” IEEE Trans. Power Del., Vol. 12, no. 2, pp. 908-915. [5] Megahed, A.I.; Monem Moussa, A.; Elrefaie, H.B.; Marghany, Y. M., "Selection of a Suitable Mother Wavelet for Analyzing Power System Fault Transients" 2008 IEEE Power and Energy Society General Meeting - Conversion and Delivery of Electrical Energy in the 21st Century, Pittsburgh, PA , pp. 1-7, July (2008). 1456 ISSN 1013-5316; CODEN: SINTE 8 [6] Waqar Adil, Aslam P. Memon, M. Usman Keerio, Ahsan Zafar, “Simulation of Power System Transient Disturbances in MATLAB”, International Journal of Emerging, Science & Engineering (IJESE), www.ijese.org, Volume. 2, Issue 3, pp. 53-58(Jan 2014). [7] Waqar A. Adil, Muhammad Usman Keerio, Aslam. P. Memon (2014), “Investigation of suitable Mother Wavelet Transform Functions for Detection of Power System Transient Disturbances”, 1st International Conference on Modern Communication & Computing Technologies (MCCT'14), 26-28 February, 2014, ISSN 2311-5297. Quaid-e-Awam University of Engineering, Science & Technology, Nawabshah, Sindh, Pakistan. http://mcct.quest.edu.pk/submission.html [8] J. Chen, W. Kinsner, and B. Huang, "Power System Transient Modeling and Classification" in Proc. IEEE 2002 Can. Conf. Electrical and Computer Eng., ISBN: 0-7803-75149; vol. 1 of 3, pp. 184-189(2002). [9] M. Karimi, H. Mokhtari, and M. R. Iravani, (2000) "Wavelet based On-line Disturbance Detection for Power Quality Applications," IEEE Trans. on Power Delivery , vol. 15, no. 4, pp. 1212-1220. [10] C. H. Kim, H. Kim, Y- H. Ko, S. H. Byun, R.K. Aggarwal, and A.T. Johns, “ A novel fault–detection technique of high–impedance arcing faults in transmission lines using the wavelet transform” IEEE Trans. Power Del, Vol. 17, no. 4, pp. 921-928(2002). [11] L. S. Safavian, W. Kinsner, and H. Turanli, (2004) "Classification of Transients in Power Systems using Multifractal Analysis" in Proc. IEEE Can. Conf. Electrical and Computer Eng, vol. 3 of 4, pp. 14451448. [12] Safavian, L. S.; Kinsner, W.; Turanli, H., "A Quantitative Comparison of Different Mother Wavelets for Characterizing Transients in Power Systems" Electrical and Computer Engineering,. Canadian Conference on, vol. 3, issue 2, pp.1445, 1448, May (2004). [13] David C, Robertson, Octavia I. Camps, Jeffrey S. Mayer, and William B. Gish, “Wavelets and Electromagnetic Power System Transients” IEEE transaction on power Delivery Vol. 11, No.2, pp. 10501058(1996). Sci.Int.(Lahore),26(4),1447-1456,2014 [14] F. Mo and W. Kinser, “Wavelet Modeling of Transients in Power Systems” Proc. IEEE, Communication, computer & Power Conf., WESCANEX 97, pp. 132-137(1997). [15] Aslam P. Memon, M. Aslam Uqaili, Zubair A. Memon and Asif Ali Akhund, “Time-Frequency Analysis Techniques for Detection of Power System Transient Disturbances,” International Journal of Emerging Trends in Electrical and Electronics (IJETEE), IRET publication www.iret.co.in, http://www.iieee.co.in, Vol. 9, pp. 39-44 (November 2013). [16] J. C. Das, ‘Transients in Electrical Systems’, Second Edition, McGraw-Hill Publication , 2910. [17] Ma, H.; Girgis, A.A, "Identification and tracking of harmonic sources in a power system using a Kalman filter," Power Delivery, IEEE Transactions on, vol.11, no.3, pp.1659, 1665(1996). [18] Paulo F. Ribeiro, "Wavelet Transform: An Advanced Tool for Analyzing Non-Stationary Harmonic Distortions in Power Systems" Proceeding of IEEE International Conference on Harmonics in Power Systems, Bologna, Italy, pp. 365-369(1994). [19] John Williams and Kevin Amaratunga, "Introduction to Wavelets in Engineering" International Journal for Numerical Methods in Engineering, Vo1.37. pp. 23652388(1994). [20] P. S. Addison. The Illustrated Wavelet Transform Handbook, IOP Publishing Ltd, ISBN 0-7503-06920(2002). [21] Ismail Y., Gulden Kokturk, “Failure Analysis in Power Systems By the Discrete Wavelet Transform” WSEAS Transaction on Power Systems, issue 3, Vol. 05, pp. 243-249(2010). [22] T. Lachman, Aslam P. M, T.R. Mohad and Zubair A. M, “Detection of Power Quality Disturbances Using Wavelet Transform Technique”, International Journal for the Advancement of Science & Arts, Vol. 1, No. 1, (2010).