Real Time Detection and Classification of Single and Multiple Power Quality Disturbance Based on Embedded S- Transform Algorithm in Labview

Real Time Detection and Classification of Single and Multiple Power Quality Disturbance Based on Embedded S- Transform Algorithm in Labview Mohd Fais Abd Ghani, Ahmad Farid Abidin and Naeem S. Hannoon Faculty of Electrical Engineering Universiti Teknologi Mara Shah Alam, Selangor, Malaysia Fais_ghani90@yahoo.com Abstract This paper present the features extraction of the real time voltage signal performed by S-Transform analysis for the purpose of detection and classification of PQ disturbance, which focus on voltage sag, voltage swell and transient. The extracted features will be used as a parameter in detecting and classifying the single and multiple PQ disturbances. As for validation purpose, the real time S-Transform based disturbance analysis is compared with Continuous Wavelet Transform. The result indicate that the S- Transform is more superior and provide more accurate analysis result especially in detection and classification of multiple PQ disturbance. The analysis of the transformed voltage signal features is conducted in LabVIEW software with the aid of data acquisition module voltage measurement to acquire voltage signal with PQ disturbance generated by Chroma Programming AC Source. In order for the S- transform to provide significant and precise S-matrix features, the real time voltage signal must be acquired with appropriate data acquisition module and adequate sampling rate. Keywords S-transform; continuous wavelet transform; voltage sag; voltage swell; transient; labview; 1. Introduction The delivery of established electrical energy power without any Power Quality (PQ) disturbance is a critical issue for the power utilities worldwide. Accurate detection of PQ disturbance is necessary for decision making and further mitigation process. Demand for instantaneous and precise detection of PQ disturbances had increase the interest in implementation of innovative PQ monitoring system with appropriate analysis algorithm and software. Most common occurred disturbances in power system line is voltage sag, voltage swell and transient which might cause problems such as malfunction, instabilities, short lifetime and equipment failure(bollen 2001),(Institute of Electrical and Electronics Engineers 2009). Various signal processing techniques have been implemented in order to transform and extract all possible features from the raw voltage signal containing PQ disturbances. These features will be used to recognize type of PQ disturbances lies within the acquired voltage signal. Short Time Fourier Transform (STFT) is one of the time analysis algorithms which widely used tools in detecting and classifying of PQ disturbance. However, STFT applies constant window size which is not appropriate for non-stationary signal(abozaed et. al 2013)(Kapoor et. al 2015)(Srividya et. al 2013)(M.F Faisal et. al 2009). Due to the certain limitation of STFT, Wavelet Transform (WT) is introduced in PQ monitoring. Unlike STFT, WT window size varied according to the signal, that applies narrow and wide window size for high and low respectively. Hence grant better time and resolution for the non-stationary signal. WT can be computed through Continuous Wavelet Transform (CWT) (Abozaed et. al 2013)(Srividya et. al 2013)(Reddy et. 280

al 2014). The aforementioned characteristic of the CWT is applicable in detecting the PQ disturbance such as voltage sag, swell and transient. Paper (Fais et al. 2016) shows the significant of real time voltage sag event detection using CWT. However, features provide by CWT is insufficient for the process of classifying PQ disturbances. Moreover, this technique also fails to differentiate multiple disturbances occur simultaneously during the monitoring work(kapoor ey. al 2015). In order to overcome those limitations, S-Transform (ST) is used to provide superior features extracting approach compared to CWT and STFT. ST algorithm applies windows that varies inversely to the present in the signal which is similar with CWT but retain absolutely referenced phase information. ST provide S-matrix representation of a real time voltage signal which is identical to the STFT(Stockwell et al. 1996). In this research, real time features extraction for the multiple PQ disturbances within the acquired voltage signal is perform based on S-transform (ST) technique. Features extracted in term of two dimensional S-matrix from ST provide better information in detecting and classifying multiple PQ disturbance compared to inadequate features of CWT coefficient, which is only applicable in detecting single PQ disturbance. This research is conducted in real time environment, where is the entire data acquisition module, analysis module, features extraction process and data representation is develop and executed in LabVIEW software. Along with huge flexibility in interfacing and acquiring various type of real time input signal from different type of measurement tools, LabVIEW is offering a high quality analysis with a variety choice of programming tools which is suitable to be implement in real time power quality monitoring system. This superior flexibility of LabVIEW software also grants an excellent output result of the projects with shorter period of completion time(vento 1988). 2. Methodology 2.1 S-Transform Algorithm As a modification from CWT, ST can be derived by multiplying the phase factor to CWT equation as expressed in (1) and the algorithm of CWT can be define as (2) ST τ, f = CWT τ, d e i2πfτ dt (1) CWT τ, d = + x t ψ t τ, d dt (2) By substituting (2) into (1), the equation for the ST can be define as, ST τ, f = + x t ψ t τ, d e i2πfτ dt(3 (3) Unlike CWT, ST applies Gaussian window instead of mother wavelet which is can be defined, ψ t, f = g(t)e i2πfτ (4) Where g(t) can be expressed as, g t = 1 σ 2π e t2 2σ (5) The capability of the ST Gaussian window to varies according to the of the signal is depend on the Gaussian window width, σ which can be define as, σ f = 1 f (6) Hence, the final equation of the ST can be expressed as, ST τ, f = f 2π + x t e (t τ)2 f 2 2 e i2πfτ dt (7) 281

2.2 Features Extraction Using S-Transform The output of the ST on the acquired real time voltage signal is an N M matrix called S-matrix. Each element of the S-matrix is present in the form of complex value. The rows of the S-matrix related to the and the column pertain to time. Hence, the output of the S-matrix can be representing by time-amplitude graph and -amplitude graph, where the amplitude stands for the absolute value or vector of each complex element from the S-matrix. Figure 1 shows the acquired real time voltage signal containing voltage sag event rated at 230V and 50Hz. As shown in figure 2, time-amplitude graph indicates the resultant S-matrix vector at a specific time corresponding to the time of input voltage signal. The transpose or inverse of the S-matrix vector can be represent as amplitude graph as shown figure 3, which express the resultant S-matrix vector corresponding to. Figure 1. Acquired real time voltage signal generated by Chroma Programming Figure 2. 2D array time-amplitude graph of S-matrix vector Figure 3. 2D array -amplitude graph of S-matrix vector Based on the S-matrix vector extracted from the transformed signal, the maximum value of each column of the vector in both 2D array time-amplitude and -amplitude graph is obtained and illustrated as shown in figure 4-6. Based on figure 4, the maximum value of each column of the -amplitude graph in figure 3 gives information regarding the fundamental and non-fundamental presents within the signal based on total number of peak. In figure 5, the maximum value of the time-amplitude s-matrix vector gives information regarding the amplitude value along the fundamental with corresponding to time. Meanwhile, in figure 6, the maximum value of the time-amplitude s-matrix vector gives information regarding the amplitude value along the non-fundamental with corresponding to time. Due to the absent of non-fundamental in voltage sag event, the time-amplitude graph of figure 6 represent almost zero value. These parameters are referred in recognizing the types of PQ disturbance discovered within the acquired voltage signal. Figure 4. Maximum value of s-matrix vector in -amplitude graph 282

Figure 5. Maximum value of s-matrix vector in time-amplitude graph index at fundamental Figure 6. Maximum value of s-matrix vector in time-amplitude graph index at non-fundamental range 2.3 S-Transform Extracted Features Characteristic All features collected from previous section approach are used to recognize the types of PQ disturbance. The detection and classification of PQ disturbance focus on the two types of short time PQ disturbance which is voltage sag, voltage swell and transient which is possibly to take place at a same period of time. The block diagram for the PQ disturbance detection and classification is illustrated in figure 7. In order to provide significant classification approach, all features characteristic based on S-matrix output can be describe as follows: Features 1: Fundamental and non-fundamental peak based on maximum value of S-matrix vector in -amplitude graph Features 2: Amplitude value based on maximum value of S-matrix vector index at fundamental Features 3: Zero and non-zero value based on maximum value of S-matrix vector index at non-fundamental Acquired Non- Stationary Voltage Signal Signal Processing Using S- Transform Features Extraction output represent as S-matrix Voltage Sag & Swell without =1 Total No. of Frequency peak Present in time graph >1 Zero and non-zero value based on maximum value of S- matrix vector index at non-fundamental Amplitude value based on maximum value of S- matrix vector index at fundamental Voltage Sag Voltage Swell Amplitude value based on maximum value of S-matrix vector index at fundamental Voltage Sag & Voltage Swell & only Figure 7. Classification stages of single and multiple PQ disturbances 283

The classification of single and multiple PQ disturbance based on above features is listed in Table I. Table 1.Classification of PQ Disturbances Based on S-matrix Features PQ Features Disturbance 1 2 3 Voltage Sag One Minimum amplitude Zero value value lower than 90% detected peak of nominal value Voltage Swell Voltage Sag and Voltage Swell and One peak More than one peak More than one peak More than one peak Maximum amplitude value higher than 110% of nominal value No change in amplitude value. Minimum amplitude value lower than 90% of nominal value Maximum amplitude value higher than 110% of nominal value 2.4 Experiment Overview and Real-Time S-Transform Analysis Setup Zero value detected Non zero value detected Non zero value detected Non zero value detected In order to comprehend the implementation of ST in real-time PQ analysis, those PQ disturbances is generated by 1.5kVA Chroma Programming AC source. IEEE-1159 standard is considered while configuring the output of single and multiple PQ disturbances such as voltage sag, voltage swell, transient, voltage sag with transient and voltage swell with transient. Voltage measurement module interfaced with user laptop through Ethernet is used to acquire real time voltage signal from AC source at 3kS/s sampling rate. To perform the ST analysis on the real time signal, the algorithm for the ST developed in the Matlab is embedded in LabVIEW software. The embedded process of the ST algorithm is realized via Mathscript Node that executes the ST algorithm created in LabVIEW Mathscript Window. The block diagram for entire experiment procedure is shown in figure 8. Generating single phase voltage signal with PQ event from Chroma programming 300 Vrms Data Acquisition Module used to acquired 230V real time voltage signal from Chroma Data Acquisition Chassis slotted with 300 Vrms for interfacing with Labview Software through Ethernet Data logging for Evaluation and Comparison Real Time S-Transform Features Extraction S-Transform Analysis on the real time signal in LabVIEW Voltage Signal from Voltage Measurement Device Continuously Acquired using DAQmx Visual Instrument Figure 8. Hardware and software implementation for real time ST analysis block diagram 284

3. Result 3.1 Verification of Multiple PQ Disturbance Detection and Classification using ST and CWT Algorithm on IEEE Recorded Waveform In order to verify the compatibility and accuracy of ST algorithm in analyzing multiple PQ disturbance events, recorded waveform from IEEE 1159.2 Working Group is used for the analysis and verification purpose. Recorded waveform containing multiple PQ disturbances, voltage sag and transient is shown in figure 9. Figure 10, figure 11 and figure 12 show the features extracted using ST in form of 2D array time-amplitude and -amplitude s- graph matrix respectively. The detection of voltage sag and transient event within the analyze voltage signal is show in figure 12 and 13. Figure 9. RecordedVoltage sag event containing transient from IEEE 1152 Working Group Test Waveform Figure 10. ST time-amplitude s-matrix for voltage sag and transient Figure 11. ST -amplitude s-matrix for voltage sag and transient Figure 12. ST -amplitude s-matrix for voltage sag and transient by observing from 100Hz and above Figure 13. Detection of voltage sag event Figure 14. Detection of transient event 3.2 Comparison of Real Time Multiple PQ Disturbance Detection and Classification using ST and CWT In detection of PQ disturbance, CWT coefficient is used to indicate the occurrences of voltage sag, voltage swell and transient event within the signal. Although CWT has the capabilities in detection of PQ disturbance, CWT coefficient unable to provide enough features for PQ disturbance classification. Based on ST algorithm, the feature is extracted from the s-matrix output which can be represented in time- contour. In time- contour, the event of multiple PQ disturbances can be recognized visually. Multiple PQ disturbance analysis for detection and classification is performed by CWT and ST as shown in figure 15-17. Figure 15. Real time voltage signal contains voltage sag and transient. 285

Figure 16. CWT coefficient for voltage sag and transient event. Figure 17. ST time- contour for voltage sag and transient The analysis of multiple PQ disturbances executed by both CWT and ST clearly shows that ST provide advance features and more accurate in detecting and classifying multiple PQ disturbance. As shown in figure 16, in the event of multiple PQ disturbances, CWT coefficient is unable to gives information regarding the types of PQ disturbances that occur at an identical period of time space. However, based on ST time- contour, the types of PQ disturbances can be classified based on the range. As shown in figure 17, the event of voltage sag take place in 50-60Hz range, while the transient event tends to arise in 200-500Hz range. Although the PQ disturbance such as transient is possibly to occur at a same period of the voltage sag event, ST analysis can clearly provide more accurate information in detection and classification of the multiple PQ disturbances. 3.3 Single and Multiple PQ Disturbance Classification Using ST In order to evaluate the capability ST in real time detection and classification of single and multiple PQ disturbance, features from real time PQ disturbance generated by Chroma Programming AC Source is extracted. Those generated voltage signal consists of single PQ disturbances such as voltage sag, voltage swell and transient, while multiple PQ disturbances focus on voltage sag and swell containing transient. Based on all features characteristic explain in section 2.3, single and multiple PQ disturbance can be detected and classified accordingly. Figure 18(a)-22(a) show the acquired real time voltage signal containing PQ disturbances. Figure 18(b)-22(b) represent the maximum value of -amplitude s-matrix vector. Figure 18(c)-22(c) shows the maximum value of time-amplitude s-matrix vector indexed at fundamental, 50Hz and Figure 18(d)-22(d) shows the maximum value of time-amplitude s-matrix vector indexed at maximum non-fundamental range. Figure 18. (a) Voltage sag (b) maximum value of -amplitude s-matrix vector (c) time-amplitude s-matrix vector index at 50Hz (d) timeamplitude s-matrix vector index at non-fundamental Figure 19. (a) Voltage swell (b) maximum value of -amplitude s-matrix vector (c) time-amplitude s-matrix vector index at 50Hz (d) timeamplitude s-matrix vector index at non-fundamental 286

Figure 20. (a) (b) maximum value of -amplitude s-matrix vector (c) time-amplitude s-matrix vector index at 50Hz (d) timeamplitude s-matrix vector index at non-fundamental Figure 21. (a) Voltage sag with transient (b) maximum value of -amplitude s-matrix vector (c) timeamplitude s-matrix vector index at 50Hz (d) time-amplitude s-matrix vector index at nonfundamental Figure 22. (a) Voltage swell with transient (b) maximum value of -amplitude s-matrix vector (c) timeamplitude s-matrix vector index at 50Hz (d) time-amplitude s-matrix vector index at non-fundamental The event of voltage sag and swell do not varies from the fundamental of the voltage signal which is at 50Hz. Hence, the time-amplitude value at corresponding will be analyzed in order to categorize the event. All row value along 50th column index is extracted from time-amplitude s-matrix vector and used to identify the present of voltage sag or voltage swell event, where 50 th index is representing 50Hz. All row value along 50th column index is illustrated in Figure 18(c)-22(c). Features 2 characteristic is taken into account for PQ disturbances detection and classification in figure 18(c)-22(c), where the amplitude during normal condition and sag/swell event is differentiate. The amplitude value during the voltage sag event will tends to drop more than 90% of the amplitude value during normal condition. Consequently, in voltage swell event, the amplitude value tends to rise more that 110% of the normal condition amplitude value. These percentage will be used as threshold value to indicates sag and swell events. 287

The recognition of non-fundamental present within the signal is based on features 1 characteristic in section 2.3. In detecting and classifying transient event, the maximum peak value of nonfundamental is identified based on Figure 18(b)-22(b). Any non-fundamental rise above the threshold value will be used as index value in extracting the related indexed row from time-amplitude s-matrix vector. In order to indicates the occurrence of transient disturbance, features 3 characteristic is implemented in interpreting the extracted row elements. As shown in Figure 20(d)-22(d), the occurrences transient event can be recognize based on the present of non-zero amplitude value within the signal time frame. Meanwhile, during the absent of transient event, the amplitude value maintained at zero for the entire period, which is clearly shown in figure 18(d) and figure 19(d). 4. Conclusion The attainment of ST in detecting and classifying of the real time single and multiple PQ disturbances is presented in this paper. The superiority of the features extraction by ST significantly provides important and sufficient information in classification multiple PQ, which is not obtainable from CWT algorithm. However, ST analysis provides more reliable features with high sampling rate of the acquired voltage signal. Hence, proper data acquisition module with sufficient sampling rate is implemented as a solution for this issue. Based on all extracted features from ST, multiple PQ disturbances such as voltage sag and swell associate with transient can be detected and classified effectively. Acknowledgement The author acknowledges the financial support given by Ministry of Science, Technology and Innovation (MOSTI) Malaysia for sponsoring this research in the form of grant-in-aid 100-RMI/SF 16/6/2 (5/2015). References Abozaed, Mohamed E Salem. 2013. Detection and Classification of Power Quality Disturbances Using S- Transform and Wavelet Algorithm. International Journal of Electrical, Computer, Energetic, Electronic and Communication Engineering 7(6): 772 77. Bollen, Math H.J. 2001. IEEE press series on power engineering Understanding Power Quality Problems: Voltage Sags and Interruptions. piscataway. Fais, Mohd, Abd Ghani, Ahmad Farid Abidin, and Naeem S Hannoon. 2016. Comparison of Real Time Voltage Sag Detection Based on Discrete and Continuous Wavelet Transform in LabVIEW. In 2016 IEEE Conference on Systems, Process and Control,, 16 18. Institute of Electrical and Electronics Engineers. 2009. 2009 IEEE Std 1159-2009 (Revision of IEEE Std 1159-1995) IEEE Std 1159 - IEEE Recommended Practice for Monitoring Electric Power Quality. Kapoor, Rajiv. 2015. Power Quality Disturbences Detection and Classification : A Review. Journal of basic and Applied Engineering Research 2(5): 405 14. M.F Faisal, and A. Mohamed. 2009. Identification of Multiple Power Quality Disturbance Using S-Transform and Rule Based Classification Technique. Journal of Applied Science 9(15): 2688 2700. Reddy, Maddikara Jaya Bharata, Dusmanta Kumar Mohanta, and Kondaveti Sagar. 2014. A Multifunctional Real- Time Power Quality Monitoring System Using Stockwell Transform. IET Science, Measurement & Technology 8(4): 155 69. http://digital-library.theiet.org/content/journals/10.1049/iet-smt.2013.0091. Srividya T, Mr. A. Muni Sankar, T. Devaraju. 2013. Identifying, Classifying Of Power Quality Disturbances Using Short Time Fourier Transform And S-Transform. Weekly Science 1(1): 1 6. Stockwell, R G, L Mansinha, and R P Lowe. 1996. Localization of the Complex Spectrum: The S-Transform. 44(4): 998 1001. Vento, J Anthony. 1988. Application of Labview In Higher Education Laboratories. In Frontiers In Education Conference Proceedings,, 444 47. 288

Biography Mohd Fais Abd Ghani was born in Malaysia, on August 27, 1990. He received her Diploma in Electrical and Electronic Engineering from Universiti Teknologi MARA (UiTM) Pulau Pinang, and Bachelor of (Hons) Electrical Engineering from Universiti Teknologi MARA (UiTM) Shah Alam in 2011 and 2014, respectively. He is currently pursuing his studies on M.Sc in Electrical Engineering at Universiti Teknologi MARA (UiTM) Shah Alam. His special fields of interests included signal processing and power quality. Ahmad Farid Abidin was born in Malaysia, on Dec 25, 1978. He received his Bachelor of Engineering in Electrical, Electronic, and System Engineering from Universiti Kebangsaan Malaysia (UKM), M.Sc in Electrical Engineering from Universiti Teknologi MARA (UiTM), and Ph.D degree from Universiti Kebangsaan Malaysia (UKM), in 2000, 2005, and 2011, respectively. He is currently a lecturer and Head of Centre for Electrical Power Engineering Studies (CEPES) in Universiti Teknologi MARA (UiTM) Shah Alam, Selangor. His main research interests are in power system stability, power quality, power system protection and high voltage. M. H. Naeem is a senior lecturer in the department of electrical engineering at Universiti Teknologi Mara. Dr. Naeem Hannoon has obtained his B.Sc. with Distinction in Electrical Engineering in 1994, M.Sc. from University Putra Malaysia 1998 and PhD from Multimedia University. He has over 12 years experience as academician and as an industrial Engineer. Dr Naeem has published several technical papers for International journals and conferences. He is a member of IEE and IASTED. He is also a technical reviewer at the Journal of the Franklin Institute, Elsevier. His interest is in application of soft and intelligent computing techniques to solve different power system problems. 289