A Real-Time Current Harmonic Monitoring System Based on Stockwell Transform Method

Transcription

1 International Review o Electrical Engineering (I.R.E.E.), ol. 11, N. 2 ISSN March April 2016 A Real-Time Current Harmonic Monitoring System Based on Stockwell Transorm Method Margo Pujiantara, Dimas Okky Anggriawan, Anang Tjahjono, Dein Permadi, Ardyono Priyadi, Mauridhi Hery Purnomo Abstract Harmonic analysis currently plays an important role or the power quality analysis because the non-linear loads, as source o the harmonic distortion, have signiicantly increased. The harmonic distortion inluences power quality in power systems such as overheating o electrical equipment and aging o electrical equipment. Thereore, non-linear loads should be monitored in real time. This research proposes the design and implementation o the prototype current harmonic analysis that can be applied in real time based on S-transorm Method. The equipment o this prototype can be explained as ollows: CT-235, TZ2L9 and CT or KWH meters as current sensor, microcontroller AT mega 8535 as data acquisition, and isual Basic sotware as displaying current waveorm, current spectrum, rms current value, and total harmonic distortion (THD) or current. S-transorm method is used to obtain the THD or current. The experimental result is validated by FLUKE 43B and the THD or current is accurately measured. Copyright 2016 Praise Worthy Prize S.r.l. - All rights reserved. Keywords: Real-Time Monitoring, Microcontroller, S-Transorm, THD i Nomenclature Input o waveorm voltage DC voltage rom power supply cc Output voltage to input analog o microcontroller R out R i ht ariabel resistance Feedback resistance Input data in time domain W,d Output wavelet in time-requency domain Mother wavelet G m,n Gaussian window S kt,n Output ST in time-requency domain THD Total Harmonic Distortion I. Introduction Nowadays, the harmonic analysis is a key issue or power system quality and security since residential, commercial and industrial loads are equipped with electronic devices [1]-[4]. The electronic devices such as computer, television, lamp, DC motor, Induction motor, monitor, Air Condition (AC) and etc. are the main components o current harmonic source which may increase the harmonic distortion. This harmonic distortion can decrease the power quality, it can increase temperature, power losses, malunction o measurement and inally it can decrease the aging o equipment [5]-[7]. The level o harmonic distortion according to IEEE standard shows that Recommended Practices and Requirement or harmonic control in electrical power system are reerred to the limit level o coniguration accepted in current harmonic o power system. This standard implies that the power consumers o power have an important role to keep low current harmonic level [8]. Thereore, the monitoring o harmonic distortion plays an important role or power system quality and security to maintain the power system equipment. However, the commercial current harmonic measurements have some problems and disadvantages. They are expensive and diicult to integrate with other measurements and to develop the sotware package. Though, many techniques have been used or the harmonic analysis, as Fast Fourier Transorm (FFT). FFT is successully used on stationary signals when the requency o the signal does not change over time. However, FFT is not suitable or detecting nonstationary signals such as oscillatory transient or spike [9], [10], [11]. To deal with non-stationary disturbances, short-time Fourier transorm (STFT) is required [12]. STFT Limit is not capable to track the dynamic signal properly due to the limit width o the window ixed to aect the requency-time resolution. Thereore, STFT is not successul or analyzing the signal o the oscillatory transient consisting o high and low requency components [12]. In order to mitigate these limitations, the wavelet transorm (WT) is applied [13]-[17]. WT can extract the important inormation rom signal disturbances and it can determine the signal disturbance types causing a harmonic distortion problem. Copyright 2016 Praise Worthy Prize S.r.l. - All rights reserved DOI: /iree.v11i

2 Moreover, WT can use short windows at high requencies and long windows at low requency. Hence, WT has the good time-requency resolution. ST is oten called as a modiied wavelet transorm because it is an extension to the ideas o WT. However, ST ensures the phase correction o WT. ST is based on a moving and scalable localizing Gaussian window [18]. ST generates a time-requency domain similar to STFT with a width-varying window. These eatures can signiicantly improve the detection harmonic distortion. This method can overcome the weakness o other transormations as stated beore. Thereore, in this paper, S-transorm (ST) is applied or current harmonic analysis. Many techniques in the literature [8]-[17] have been used or the harmonic analysis but they have not used real time. Thereore, the ST algorithm is proposed or real time analysis mode and the design and implementation o prototype current harmonic analysis are proposed in this research. The equipment o this prototype can be explained as ollows: CT-235, TZ2L9 and CT or KWH meters as current sensor, microcontroller AT mega 8535 as data acquisition, and isual Basic sotware as displaying current waveorm, current spectrum, root-mean-square (rms) current value, and total harmonic distortion (THD) or current. The ollowing parameters are calculated in the real time mode. The experimental results are validated by FLUKE 43B and accurately measured through the THD. The results or current with the average percentage errors are less than 3%. The most signiicant contribution o this experiment is that THD or current can be observed using a computer monitor in real time. The organization o this paper is as ollows: Section II discusses hardware design, section III describes ST, section I presents Simulation Result and Analysis, section shows Experimental Result and Analysis and section I gives the conclusion and reerences. II. Hardware Design This section explains the current harmonic detection consisting o three parts. The irst part presents the data acquisition design. In the second part there is the description o the sotware applied or the analysis the harmonic current distortion. Finally the third part shows an experiment o the prototype. Fig. 1 shows a block diagram o harmonic current monitoring consisting o PC, microcontroller AT mega 8583, digital input-output, analog input-output, and Current Transormer (CT). II.1. Data Acquisition Design Data acquisition design is connected between current transormer and microcontroller based on the speciication CT-235, TZ2L9 and CT or KWH meters and microcontroller. These speciication are obtained by testing the devices using standard measurements such as digital Oscilloscope BK-PRECISION and GDM Fig. 1. Block diagram o current harmonic monitoring Parameter, value, and inormation rom the testing devices are shown in Table I. The parameter values used or the current harmonic analysis are represented as current waveorm. The microcontroller as a device or converting analog to digital (ADC) has the ollowing speciication o analog input data: 0 5. TABLE I PARAMETER ALUES OF CURRE TRANSFORMER alue Parameter CT235 TZ2L9 CT or KWH Inormation CT ratio Ratio o CT primary and secondary values R p CT Output resistance o CT Gain Gain o data acquisition Oset Oset o data acquisition The waveorm current is lowed rom a transormer to a microcontroller with the unction o changing the waveorm current into a voltage waveorm. Moreover, the voltage waveorm is adapted into an input data analog and this current waveorm can be detected by the microcontroller. The waveorm current is processed rom the buer circuit, the inverting ampliier circuit and the inverting adder ampliier circuit. In the inverting ampliier, the input o voltage waveorm is changed into negative voltage. This voltage increases when the gain is adjusted. This happens because the voltage lows rom inverting ampliier circuit into inverting adder ampliier circuit that is powered in DC supply. This voltage is known as oset voltage. The equation or processing current waveorm (1), or inverting ampliier circuit and or inverting adder ampliier circuit is as ollows: II.2. R R out i cc Ri Ri Sotware o Harmonic Analysis The harmonic analysis sotware aims to analyze the harmonic current distortion that needs data rom the output o acquisition device. This sotware is developed (1) 194

3 by an open source program named visual basic. There is a device to communicate rom a microcontroller to this sotware to process the data. It is known as WIZNET WIZ110SR used as ETHERNET communication. This connection is better than RS 232 serial communication because WIZNET can send data aster than RS 232 serial communication. The WIZNET has higher capability to keep the data rom missing communication. The analog data are changed into digital data in the microcontroller since the data or harmonic analysis are processed in digital. The microcontroller manages the value o baud rate, number o ADC bit, and the time delay or data communication rom prototype with isual basic sotware. These processing data rom analog to digital are represented in Table II. Parameter, value, and inormation are also shown. Fig. 3. Scheme o prototype test circuit Fig. 4. Design and implementation o prototype Fig. 2. Block diagram o current harmonic analysis TABLE II PARAMETER ALUES OF ADC PROCESSING Parameter alue Inormation Baud rate kbps Rate o data transmitting Bit ADC 8 Number o bit Time delay 0.2 millisecond sampling rate Clock 1000 kbps Clock maximum For the signal conditioning o current waveorm in original terms the Eqs. (2) and (3), and are needed: input II.3. I ADC 5cc 255 input (2) 2000 (3) 200 Prototype Testing The Prototype o the current harmonic monitoring has some displayed eatures, such as current waveorm, rms current value, current spectrum and THD. Its testing perormed at a single phase load or linear and nonlinear. The result o the prototype monitoring is compared with standard measurement tools. The results rom current waveorm are rms current, current spectrum, and THD displayed in view orm. Fig. 5. Current harmonic dashboard monitoring TABLE III STANDARD MEASUREME TOOLS ARE CONDUCTED IN PROTOTYPE TESTING Parameters are Standard measured measurement tools rms current level Digital Multi meter (GDM-8145) Current waveorm Digital Oscilloscope BK precision (BK-2542) THD or current level Fluke 43B III. The Proposed Algorithm III.1. Description o ST ST is the main part o harmonic analysis sotware. It can generate the current spectrum used to obtain THD. ST is one o the harmonic analysis methods with superior eatures respect to other methods. ST equation can be obtained rom the continuous wavelet transorm (CWT). W,d h t is deined as: The CWT o a unction 1 W,d h t t dt (4) d d 195

4 ST o a unction ht can be deined as CWT with mother wavelet multiplied by phase actor and by replacing dilation d with inverse o requency, as deined below: i2 S, e W, (5) 2 The mother wavelet is deined as ollow: 2 2 t 2 i2 t t e e (6) The dilation actor d is the inverse o requency. Thereore, the inal orm o the continuous ST is obtained as ollow: t j t S, h t e e dt (7) 2 The width o the Gaussian window is: a (8) ST o a unction h t is expressed in discrete orm as hkt, k = 0,1,, N-1. T is the sampling time interval and N is the total number o sampling. ST rom discrete time series hkt is given with (τ kt and n/), which can be written as [14]-[17]: N 1 j2 mk n m n S kt, H G m,n e, n where: N 0 (9) M 0 G m,n m / n e (10) and k, m=0,1,,n-1 and n=1,., N-1. or n = 0: N 1 1 m S kt, 0 h N (11) m0 n kt, Arg S kt, n n Im S, jt =Arg n Re S, jt (12) n Matrix S kt, and phase n kt, o the S matrix are used or the harmonic analysis where the rows o the matrix represent magnitude in requency and the columns represent magnitude in time. III.2. Harmonic Calculation Harmonic distortion is caused by non-linear loads, its characterization is the periodic signal and its value can be obtained rom the amplitude multiplied by the requency representing the magnitude and the phase o harmonic component individually. ST provides the amplitude requency inormation to obtain the harmonic component. An indicator is used to show the value o the harmonic distortion in the total harmonic distortion (THD). ST gives the current amplitude in requency domain to obtain THD value and can be written as below: THD i N 2 in n2 i 1 (13) where, THD is harmonic o current, i n is current amplitude o harmonic n-th and i 1 is current amplitude o undamental requency. I. Simulation Result and Analysis ST represents the requency-time analysis known the best technique or perorming signal processing o nonstationary signals. In this section, harmonic and harmonic with sag signals are analyzed using MATLAB simulation to prove the capability o ST. The results o ST are compared by using FFT. The modeling o harmonic signal done with requency sampling o 1000 Hz, shows such results: harmonic signals appears to 3-9 occurrences at 0.2 s. The harmonic signal is analyzed by ST and Figs. 6 show the analysis results. The modeling o harmonic sag signal simulated with requency sampling o 1000 Hz o shows that harmonic signal happens at 3-9 occurrence at 0.2 s and sag occurrence rom 0.08 s to 0.12 s. The harmonic signal is analyzed by ST and Figs. 7 show the analysis results. From the Fig. 7(b), it can be seen FFT is not capable to track sag signal. In the Fig. 7(c), it is seen that ST contours show a magnitude reduction similar to a voltage sag disturbance.. Experimental Result and Analysis The proposed method is applied to perorm the measurement o parameters such as current waveorm, rms current value and THD or current. To evaluate the capability o the proposed method in the industrial power system, the commercial products such as FLUKE 43B, Digital Oscilloscope BK precision and GDM-8145 are used as the real measurements. 196

5 5 Amplitude () (a) (a) Amplitude () Frequency (Hz) (b) (b) Frequency (Hz) (c) (c) Figs. 6. (a) Harmonic signal. (b) Amplitude requency analysis o a test signal. (c) Time requency analysis o a test signal In Fig. 3, the prototype testing uses 275 W, 550 W and 875 W resistive loads to describe linier loads and 400 W, 675 W and 950 W to describe non linier loads..1. Rms Current Testing The testing is conducted with 3 types o current transormers to linear and non-linear loads compared with GDM The test with single phase A aims to check the validity and capability prototype to represent the values o rms current. The summaries o rms current testing results or linear loads and non-linear loads are shown respectively in Tables I and. Table I shows the percentage error values o prototype with 3 types o CT compared with GDM 8145 in the linear loads. For a linear load o 275 Watt, the current transormer or KWH meters gives more accurate results compared with other current transormers. Figs. 7. (a) Harmonic sag signal. (b) Amplitude requency analysis o a test signal. (c) Time requency analysis o a test signal This happens because the current transormer or KWH meters provides very minimal average percentage errors o showing that the value o current transormer or KWH meters is 1.31%. For a linear load o 550 Watt, the current transormer or KWH meters gives more accurate results compared with other current transormers because the current transormer or KWH meters provides very minimal average percentage errors and the value o current transormer or KWH meters is 0.64%. For a linear load o 825 Watt, current transormer or KWH meters shows more accurate results compared with the other current transormer because the current transormer or KWH meters provides very minimal average percentage errors and the value o current transormer or KWH meters is 0.4%. The whole results show that the 3 types o current transormers perorm an accurate measurement o rms current as the average percentage error in the 3 types o loads is lower than 2%. 197

6 Table shows the percentage error values o prototype with 3 types o CT compared with GDM 8145 in the non-linear loads. For a non-linear load o 400 Watt, the current transormer or KWH meters perorm more accurate results than other current transormers because the current transormer or KWH meters provide very small average percentage errors and the value o current transormer or KWH meters is 1.86%. For a non-linear load o 675 Watt, the current transormer or KWH meters perorms more accurate results than the other current transormers because the current transormer or KWH meters provides very small average percentage errors and the value o current transormer or KWH meters is 0.99%. For a non-linear load o 950 Watt, the current transormer or KWH meters perorms more accurate than the other current transormers because the current transormer or KWH meters provides very small average percentage errors and the value o current transormer or KWH meters is 0.72%. The obtained results show that the 3 types o current transormers perorm accurate measurements o rms current. The average percentage error in the 3 types o loads is lower than 3%..2. Current Waveorm Testing The result o the current waveorm testing is shown in Fig. 8. The current waveorm testing is very important or checking the prototype validity and the capability o representation. The harmonic current analysis is calculated by using the current waveorm. Testing is conducted by comparing the current waveorm result o prototype with the waveorm result o oscilloscope BK From the Fig. 8, it is observed that the current waveorm o prototype line is more suitable than the oscilloscope BK-254 line. Consequently the prototype aects the rms current and making the THD more accurate. TABLE I SUMMARY OF THE RELATIE ERROR RESULTS OF PROTOTYPE AT RMS CURRE LEEL TESTING FOR LINEAR LOADS WITH 3 TYPES OF CT No Load (Watt) Average percentage error (%) Micro Precision Current CT-235 TZ2L9 Transormer or KWH Meters average TABLE SUMMARY OF THE RELATIE ERROR RESULTS OF PROTOTYPE AT RMS CURRE LEEL TESTING FOR NON-LINEAR LOADS WITH 3 TYPES OF CT No Load (Watt) CT-235 Average percentage error (%) Micro Precision Current TZ2L9 Transormer or KWH Meters average Fig. 8. Current waveorm graphic during the period.3. THD o Current Testing The THD o current tested or linear loads and nonlinear loads is used to check the validity and capability o prototype. In Table I, the proposed algorithm is compared by FFT with FLUKE 43B as reerence. For linear loads, the percentage errors o THD are 0.6% and 0.6%, respectively. For non-linear loads, the percentage errors o THD are 0.3% and 2.1%, respectively. The average percentage errors o THD are less than 3%. This means that the proposed method demonstrates an accurate application. TABLE I THE COMPARISON RESULT OF THD ALUES BETWEEN PROTOTYPE, FLUKE 43B AND FFT METHOD Type Watt THD Error (%) FLUKE 43B ST FFT ST FFT Linear Linear Non-linear Non-linear I. Conclusion This paper proposes design and implementation o prototype harmonic distortion o current based on ST. This prototype is enough accurate and capable to measure and monitor rms value, waveorm, spectrum, and THD current in the real-time system. The prototype accuracy is high, around 3% or all the measurements. This result is calibrated and validated by using some measurements such as FLUKE 43B, GDM 8145, and digital multi-meter BK- PRECISION. Reerences [1] T. Yang, H. Pen, D. Wang, Z. Wang, Harmonic Analysis in Integrated Energy System Based on Compressed Sensing, Applied Energy, vol. 165, March 2016, pp [2] Parithimar Kalaignan, T., Senthilkumar, J., Suresh, Y., A novel dissociated current control technique or harmonic minimization in non-linear loads, (2014) International Review on Modelling and Simulations (IREMOS), 7 (1), pp [3] Jayachandran, J., Sachithanandam, R.M., Artiicial intelligence based controller or series and shunt active ilters or power quality improvement, (2015) International Review o Automatic Control (IREACO), 8 (3), pp [4] Sridevi, T., Ramesh Reddy, K., Harmonic mitigation and 198

7 comparison in a 15-bus network with combined hybrid active ilter using SRF and pq-algorithms, (2014) International Review o Automatic Control (IREACO), 7 (2), pp [5] Arrillaga, J., Watson, N.R., Power System Harmonics Second Edition, John Wiley & Sons Ltd, England, Ch. 1, 2, 5, 2003 [6] D. Srinivasan, W. S. Ng, and A. C. Liew, Neural network-based signature recognition or harmonic source identiication, IEEE Trans. Power Del., vol. 21, no. 01, pp , January [7] C. F. Nascimento, A. A. Oliveira Jr., A. Goedtel., A. B. Dietrich., Harmonic distortion monitoring or nonlinear loads using neuralnetwork-method, Applied Sot Computing 13 (2013) [8] IEEE Standard , Recommended Practices and Requirements or Harmonic Control in Electrical Power Systems, Institute o Electrical and Electronics Engineers, June 1992 [9] F. Zhang, Z. Geng, W. Yuan, The Algorithm o interpolating Windowed FFT or harmonic Analysis o electric Power System, IEEE Trans. Power Del., ol. 16, No. 2, Apr [10] H.Qian, R. Zhao, T. Chen, Interhamonics Analysis Based on Interpolating Windowed FFT Algorithm IEEE Trans. Power Del, ol. 22, no. 2, Apr [11] Granados-Lieberman, D., Romero-Tronsoso, R.J., Osomio-Rios, R.A., Garcia-Perez, A., Cabal-Yepez, E., Techniques and Methodologies or Power Quality Analysis and Disturbances Classiication in Power Systems: a review, IET Generation, Transmission and Distribution, 2010 [12] Nath, S., P. Sinha, S. K. Goswami, A Wavelet based Novel Method or the detection o harmonic sources in power systems, Electrical Power and Energy Systems 40 (2012) [13] M. A. S. Masoum. S. Jamali, N. Ghaarzadeh, Detection and Classiication o Power Quality Disturbances Using Discrete Wavelet Transorm and Wavelet Networks, IET Sci. Meas. Technol. vol. 4, Iss. 4, pp , 2010 [14] O. Poisson, P. Rioual, M. Meunier, Detection and Measurement o Power Quality Disturbances Using Wavelet Transorm, IEEE Trans. Power Del. vol. 15, no. 3, Jul [15] S. Santoso, E. J. Powers, W. M. Grady, A.C. Parsons, Power Quality Disturbances Waveorm Recognition Using Wavelet Based Naural Classiier-Part 1: Theoretical Foundation, IEEE Trans. Power Del. vol. 15, no. 1, Jan [16] Ghaemi, A. H, Abyaneh HA, Mazlumi K. Harmonic indices assessment by wavelet transorm Electr Power Energy Syst 2011; 33: [17] C.H. Lin, C.H. Wang, Adaptive Wavelet Networks or Power- Quality Detection and Discrimination in a Power System, IEEE Trans. Power Del. vol. 21, no. 3, Jul [18] Stockwell, R.G., Why Use S-Transorm, Northwest Research Associates, Colorado Research Associates Division, 3380 Mitchell Lane, Boulder Colorado USA Anang Tjahjono was born in Ponorogo, East Java Province Indonesia on November 19, 1964, this time as a doctoral student at the Institut Teknologi Sepuluh Nopember. Starting in 1990 as a lecturer in Electronic Engineering Polytechnic Institute o Surabaya Indonesia, teaching ield o industrial automation, artiicial intelligence, as well as microcontroller, experienced as JICA Expert in Rwanda Arica. Research is being done at this time is the ield o adaptive protection relay in the distribution system using Artiicial Intelligent method, Currently pursuing a doctoral program in ITS. Received a bachelor's degree in 1990 in the course o electronics in ITS, and got a master's degree in control systems in ITS in Dein Permadi received the B.E. degree rom electrical engineering department, Institut Teknologi Sepuluh Nopember in He is currently as Engineer in PT. PLN. His research interests include power quality, protection coordination and artiicial intelligent. Ardyono Priyadi was born in Nganjuk east java o Indonesia on September 27, He received his bachelor degree in Electrical Power System Engineering rom Institut Teknologi Sepuluh Nopember (ITS), Indonesia in 1997, master and Ph.D. degree in Electrical Power System Engineering rom Hiroshima University, Japan in 2008 and He is currently a lecturer at Electrical Engineering Department, ITS. His research interest is power transient stability, renewable energy, and identiication o power systems. Mauridhi Hery Purnomo was born in Bangkalan east java o Indonesia on September 16, He received his bachelor's degree rom Electrical Engineering Institute Technology Sepuluh Nopember, Surabaya in 1984 and master's degree rom Electrical Engineering Osaka city University, Japan in 1989, and Ph.D. degree in Power system rom Osaka city University, Japan in Where he is currently an head Laboratory o Instrumentation, Measurement and Power System Identiication and a proessor o Electrical Engineering, Institut Teknologi Sepuluh Nopember, Surabaya. His research interest is Artiicial Intelligent, Neural Network, Image Processing, renewable energy, condition and monitoring system. Authors inormation Margo Pujiantara was born in Pasuruan east java o Indonesia on March 18, He received his bachelor degree in Electrical Power System Engineering rom Institut Teknologi Sepuluh Nopember, Indonesia in 1985, master degree rom Institut Teknologi Bandung (ITB), Indonesia in 1995 and Ph.D. degree rom Institut Teknologi Sepuluh November, Indonesia in He is currently a lecturer at Electrical Engineering Department, Institut Teknologi Sepuluh Nopember. His research interest is renewable energy, protection and identiication o power systems. Dimas Okky Anggriawan received the B.E. degree rom electrical engineering department, Institut Teknologi Sepuluh Nopember in 2013 and M. Eng degree rom electrical engineering department, Institut Teknologi Sepuluh Nopember in He is currently as lecturer in the Politeknik Elektronika Negeri Surabaya. His research interests include power quality, protection coordination and artiicial intelligent. 199