Research on Local Mean Decomposition Algorithms in Harmonic and Voltage Flicker Detection of Microgrid

Transcription

1 Sensors & Trnsducers 23 by IFSA Reserch on Locl Men Decomposition Algorithms in Hrmonic nd Voltge Flicer Detection of Microgrid Wensi CAO, Linfei LIU School of Electric Power, North Chin University of Wter Resources nd Electric Power, Zhengzhou, 4545, Chin Tel.: E-mil: Received: 23 September 23 /Accepted: 25 October 23 /Published: 3 November 23 Abstrct: In llusion to hrmonic nd voltge flicer in microgrid, the locl men decomposition lgorithm is dopted to nlyze hrmonic nd voltge flicer in power system. Complex originl signls cn be decomposed into number of PF (product function) component, ech lyer of PF is composed of the envelope signl nd frequency modultion function, which contins ll the instntneous mplitude nd instntneous frequency informtion. Further combintions cn get the originl signl time frequency distribution. Using the LMD to detect the hrmonic signl nd the multiple frequency voltge flicer signl in microgrid, The simultion results show tht this Algorithm cn dptively decompose the signl nd highlight the locl chrcteristics of PF dt, the method cn be ccurte nlysis of multi frequency hrmonic distortion signl, interhrmonic signl nd multiple frequency voltge flicer signl. Simultion wveform is not only influenced by "end effect" of smll effect, nd the instntneous frequency is lwys positive. According to the ctul nlysis of trnsformer with multi-frequency signl power in microgrid system, using LMD lgorithm nd HHT lgorithm, The result further prove the correctness of the proposed method, which provides the theoreticl fundmentl in new wy for the electricl energy detection in power system. Copyright 23 IFSA. Keywords: Locl men decomposition, Hrmonic, Voltge flicer, Power qulity detection, Microgrid.. Introduction The development of new energy nd renewble energy is n inevitble choice to solve the contrdiction between Chin's energy shortge, energy utiliztion nd environmentl protection. Renewble energy, including wind energy, solr energy, ocen energy, biomss, geotherml-bsed distributed micro grid technology is one of the development directions of the electric power industry in the 2st century. Microgrid (microgrid, MG): distributed power, energy storge devices, energy conversion device, lod nd monitoring, protection device pool from genertion nd distribution system, nd this utonomous system is ble to chieve selfcontrol, protection nd mngement [-2]. With the expnsion of the depth nd rnge of pplictions of microgrid wor reserch, micro-grid power qulity problems is getting more nd more people s ttention. Hrmonics cn be divided into stedy-stte hrmonic nd time-vrying hrmonic, nd there re mny studies for the stedy-stte hrmonic detection method [3-4]. Microgrid hrmonic disturbnce signl nd inter-hrmonic signl is non-liner, non-sttionry signls. Fourier trnsform cn not hndle non-liner, non-sttionry signls, spectrl lege nd fence phenomen lso exist between tretments hrmonic 384 Article number P_57

2 shortcomings [5-7]; Anlysis of non-liner, nonsttionry signls wvelet theory hs mny limittions, must construct divide strict nd energy concentrtion wvelet bsis; HHT power qulity detection methods chieved good results, but the decomposition of the modl experience using cubic spline interpoltion fitting the envelope signl is esy to pper envelope, owe envelope phenomenon; HHT in the excessive number of "screening" led to the end effect of pollution throughout the dt segment nd the instntneous frequency bsed HHT time frequency nlysis methods often pper to be negtive is physicl phenomen which is difficult to explin [8-9]. Jonthn Smith proposed new dptive timefrequency nlysis method locl men decomposition (locl men decomposition LMD) in 25 []. LMD time-frequency nlysis method is to decompose the originl signl into series of product function group (the Product Function, PF), the lyers PF by the envelope signl nd pure FM signl is composed of two prts, they contin ll of the instntneous mplitude nd instntneous frequency informtion, the further combintion when you cn get the originl signl frequency distribution. The LMD method hs been successfully pplied to the detection of EEG, the instntneous frequency of the signl extrction nd mechnicl fult dignosis [-]. This is the first time locl men decomposition (LMD) pplied to the power system micro grid hrmonic disturbnce signl nd inter-hrmonic signl detection, by simultion experiment verify the effectiveness of the method. 2. Locl Men Decomposition Principle 2.. Locl Men Decomposition Algorithm Locl men decomposition cn decompose ny complicted signl into number of the PF component which hs certin physicl mening nd, ech PF component by the plin envelope signl nd FM signl integrted. For signl x(, the decomposition step is s follows [3-4]: ) To determine the signl x( of ll locl extreme point n i. b) Through ech extremum point n i, clculte ny two djcent locl extreme point men m i nd the vlue of the envelope estimte i m i =(n i + n i+ )/2 () i = n i -n i+ /2 (2) Connect the djcent locl men point mi nd m i+ with broen line, nd then conduct smooth hndling by using the moving verge method to get the locl men function m (. Connect ech djcent envelope estimte vlues i nd i+ with broen line, nd then conduct smooth hndling by using the moving verge method to get the envelope estimte function (. c) Seprte the locl men function m ( from the originl signl x(, nd obtin the signl H (: h (=x(-m ( (3) d) Divide h ( by the envelope estimte function (, Get the FM signl s (: s (=h (/ ( (4) Determine whether the s ( is pure FM signl, the determintion condition is to repet the bove steps for s (, get the envelope estimtion function ( stisfies the (=, nd if not stisfy described nd s ( is not pure FM signl nd then repet n times until s ( is pure FM signl, i.e. s ( the envelope of the estimtion functions stisfy (n +) (=, so: h h h s s s = x( m = s = s m = h = h = h m ( n ) Conditions for itertive terminted: (5) (6) lim = (7) n In prcticl ppliction, in order to void excessive decomposition number, we cn set disturbnce, the itertion will end when - ( +. e) Multiply the itertive process envelope estimtion function, get the envelope signl (: n ( = t = = ( ) (8) f) Obtin the formul (8) in the envelope signl ( nd pure FM signl s ( multiplied, to obtin the originl signl x(, s PF component: PF (= ( s ( (9) The first PF component contins the highest frequency component of the originl signl. g) Seprte PF(T) from the originl signl x( to get u ( s new dt to repet the bove steps, the cycle times until u ( is monotonic function so fr. 385

3 u = x( PF u2( = u( PF2 u = u ( PF ( t ) () As cn be seen from the bove steps, the originl signl cn be reconstructed by u ( nd ll PF components, i.e.: x( = = i PF i + u () Fig.. Structure of typicl micro-grid Bsed on the Instntneous Frequency of the LMD Strie By formul (), the signl is decomposed into number of PF component nd, ech PF components represented by the pure envelope signl ( nd pure FM function s(=cosφ(, its frequency f cn be pure FM function s( directly solve, nmely: φ(=rcos(s() () Expnd the formul () nd the derivtive cn be clculted the instntneous frequency of s(, the corresponding component of the instntneous frequency of the PF. s( vlues between ±, if s( vlue is pproximtely equl to ±, ± insted becuse it is derived by the derivtive of the cosine function of the instntneous frequency of the PF. This method of obtining frequency is intuitive nd simple, referred to s "direct method", nd compred to the method of the instntneous frequency of HHT trnsform strie to strie the instntneous frequency of the "direct method" is lwys positive vlue, does not pper HHT negtive frequencies phenomenon. 3. Detection nd Anlysis Bsed on the LMD Microgrid Hrmonics nd Voltge Flicer Signl 3.. Cuses of Micro-Grid Hrmonics nd Voltge Flicer Fig. is typicl micro-grid structure digrm, photovoltic, fuel cell nd micro turbine power electronics connected to the lternting current (AC) busbr through interfce micro networ, smll wind turbine is directly connected to the micro grid. Micronetwor system uses hierrchicl control strtegy, nd llows micro-networ s prt of the distributed power grid supply to lrge power grids. Micro grid through the point of common coupling (point of common coupling PCC) connects the min grid. Hrmonics generted in the micro grid is hrmonic of the generted hrmonics nd micronetwor within the power electronic components by the distributed power genertion system includes mny distributed power itself. Micro-grids contin nonliner lod nd lot of power electronic equipment, these deflector chrged with the importnt ts of lod switching nd power trnsfer, but its opertion will cuse the grid voltge nd current wveform distortion, cuse hrmonics nd inter-hrmonics pollution. Nonliner lod current of mplitude, phse, nd wveform chnges, nd contins lrge number of hrmonics nd interhrmonics ingredients. For wind turbines, the voltility of the fns to contribute Fns to contribute due to chnges in wind speed nd wind turbine tower shdow, the fluctution precisely in to be ble to generte the voltge flicer the frequency rnge (below 25 Hz). Power trnsient disturbnce nd continued voltility re esily cused, resulting in voltge fluctution nd flicer. Motor, rolling mill, electric rc furnce, electric locomotives belong impct lod, t run time, such s the lod will me the grid voltge is unstble, resulting in slow or rpid voltge fluctution. Such impct lod chrcteristics vry, thereby generting the flicer lso vries Micro-Grid Hrmonic Signl Detection Anlysis the PCC point hrmonic current signl x( of the micro grid [5]. Hrmonic of PCC point nd inter-hrmonic current signl mthemticl expression s follows: 2sin(πt ) +.2 sin(22π, t <.2 2sin(πt ) +.7 sin(48π,.2 t <.3 2sin(πt ) +.5sin(8π,.3 t <.4 (3) The current wveform of the signl x( is shown in Fig. 2, the smpling frequency is 32 Hz, disturbnce quntity =.. To verify the dvntges of the LMD, the endpoint is not doing the processing; The LMD correltion wveform nlysis is shown in Fig

4 x(/v flicer, usully the voltge fluctutions in the frequency rnge.~35 Hz. Voltge flicer signl cn be expressed s: S = m= () t A + A cos( ω cos( ω m m, (4) PF frequency/hz (/V Fig. 2. Micro-grid time-vrying inter hrmonic signl X:.34 Y: X:.2497 Y: 24. X:.3525 Y: () Inter hrmonic signl frequency. X:.929 Y:.92 X:.25 Y:.6947 X:.349 Y: (b) Inter hrmonic signl mplitude. Fig. 3. Inter-hrmonic time frequency signl nlysis in microgrid bsed on LMD. Fig. 3() shows tht the inter-hrmonic frequencies in the -.2 s Hz,.2-.3 s s 24 Hz,.3-.4 s for 4 Hz; Fig. 3(b) shows tht the hrmonic mplitude -.2 s.2 V,.2-.3 s to.7 V,.3-.4 s for.5 V. And compre literture [23] the HHT obtined experimentl wveform cn find tht even if the endpoint LMD nlysis inter hrmonic is not be processed, the decomposition of the wveform end effect is smll, the reson LMD decomposition signl verge slip fitting envelope, void HHT cubic interpoltion prone to over-the envelope or owe envelope phenomenon; Minus the envelope function "screening" thn n intrinsic mode process nd HHT, LMD get PF component divided by the envelope function "screening". The number is significntly lrger thn the HHT get n intrinsic mode "screening" the number of components is less "Screening" the fewer the number, the less light the endpoint effect inwrd pollution dt Microgrid Voltge Flicer Signl Detection Microgrid with wind turbines nd impct lods such s motors, rolling mill, electric rc furnce, electric locomotive, will cuse the micro grid voltge where A is the frequency fundmentl voltge mplitude; A m constitute the mplitude of the hrmonics of the flicer signl; ω frequency ngulr frequency; ω m to constitute the ngulr frequency of the flicer signl hrmonics, v() t = A m cos( ω is m= flicer envelope signl. Fig. 4() is multifrequency voltge flicer wveform of signl, the corresponding expression is: S = ( +.2 cos(πt ) + cos(2πt ) + 2cos(4πt + π 6))cos(π m (5) This rticle through strie signl mximum point of the envelope of v(, then its LMD nlysis, the smpling frequency of 32 Hz, the frequency nd mplitude of the flicer signl finlly obtined. LMD decomposition of the PF, PF2 nd PF3 re respectively, the PF component LMD decomposition is rrnged in descending order of the locl frequency, Fig. 4(c) shows tht, the microgrid voltge flicer signl frequency is 2 Hz, Hz, 5 Hz, Fig. 4(d) shows tht, the mplitude of micro grid voltge flicer signl re 2 V, V,.2 V, cn be seen tht the LMD decomposition cn be efficiently extrcted out of the multi-frequency voltge in the output voltge of flicer signl, the successful reliztion of the disturbnce signl component of the frequency nd mplitude of the decomposition, exploded wveform cn be found from Fig. 4 the instntneous frequency obtined by the LMD method is very ccurte, nd strie the frequency is positive, continuous, hving physicl mening of time vrying frequency, no negtive frequency unexplined. x(/v Fig. 4(). Voltge flicer time-frequency signl nlysis in microgrid bsed on LMD multifrequency voltge flicer signl. 387

5 v(/v 6 4 fundmentl hrmonic nd inter-hrmonic. In ddition, under the sme conditions, the LMD lgorithm running s, HHT running s, the running speed of LMD is gretly improved Fig. 4(b). Voltge flicer time-frequency signl nlysis in microgrid bsed on LMD envelope signl of the voltge flicer signl. frequency/hz PF PF2 PF Fig. 4(c). Voltge flicer time-frequency signl nlysis in microgrid bsed on LMD frequency. () PF component mplitude/v PF PF2 PF Fig. 4. Voltge flicer time-frequency signl nlysis in microgrid bsed on LMD mplitude function of PF component. (b) Instntneous frequency 4. Experimentl Verifiction Fig. 5() shows, x( is multiple frequency hrmonic distortion signl of power trnsformer in microgrid system. Smpling frequency is 32 Hz, =., to filter the high frequency noise signl, the PF component using LMD lgorithm seprted s shown in Fig. 5(), the frequency were seprted from high to low order using the LMD lgorithm; the corresponding instntneous mplitude function wveform nd instntneous frequency function wveform is shown in Fig. 5(b), 5(c) bsed on HHT nd LMD. Compred with HHT lgorithm, simultion results show tht LMD lgorithm is better thn HHT lgorithm in the prmeter fluctution of trnsient chrcteristic prmeter detection, the detection ccurcy nd the end effect. The stedy frequency of ech PF component by the lest squres fitting were 25.6 Hz, 6.5 Hz, 5.4 Hz nd 25. Hz, the multi frequency hrmonic signl in the trnsformer with 5 th hrmonic, (c) Instntneous mplitude Fig. 5. Anlysis results of the multi-frequency hrmonic signl in trnsformer. 5. Conclusions Micro-grids contin lrge number of power electronics components nd shoc loding hrmonics nd voltge flicer. This rticle is the first time to use the LMD method of microgrid muttions, nonsmooth inter-hrmonic signl nd multi-frequency voltge flicer hrmonic signl nlysis, this method cn be seen from the simultion results dptive signl grdul decomposition, the PF component obtined in the decomposition projecting the locl chrcteristics of the dt, nd cn effectively determine the micro-grid disturbnce occurred nd 388

6 the recovery time s well s the mplitude nd frequency of the disturbnce signl. And the resulting wveform is less ffected by "end effect", becuse the integrl clcultion is not required, hving smll mount of computtion, speed, etc., the more importnt is the instntneous frequency of the strie with the LMD method re positive. Vlid theoreticl bsis nd new pproch for power qulity detection re supplied. Acnowledgment This wor is supported by the Nturl Science Foundtion of the Eduction Deprtment of Henn Province (A475). References []. X. Shi, W. Zho, Y. Shen, Automtic license plte recognition system bsed on color imge processing, Lecture Notes on Computer Science, Vol. 3483, 25, pp [2]. S. C. Lin, C. T. Chen, Reconstructing vehicle license plte imge from low resolution imges using nonuniform interpoltion method, Interntionl Journl of Imge Processing, Vol., Issue 2, 28, pp [3]. Jifeng Wen, Pei Liu, A new method for detection of power qulity disturbnces, Proceedings of the CSEE, Vol. 22, Issue, 22, pp [4]. Song Hn, Guo-Yue Qiu, Xio-Jun Peng, An investigtion on quntittive evlution nd detection of instntneous voltge flicer, Power System Protection nd Control, Vol. 36, Issue 6, 28, pp [5]. De-Li Liu, Yn-Bin Qu, Appliction of improved HHT pproch to hrmonic nlysis in power system, Power System Protection nd Control, Vol. 4, Issue 6, 2, pp [6]. Cun-Xing Yng, Zhn-Ying Tong, Yu-Ho Wn, et l., Anlysis nd reserch of the trnsient composite disturbnce signl of power system bsed on HHT, Power System Protection nd Control, Vol. 37, Issue, 29, pp [7]. Wei-Li Bi, Zhi-Gng Liu, Qun-Wei Peng, et l., Reserch of the lod forecsting model bsed on HHT nd combintion of ANN, Power System Protection nd Control, Vol. 37, Issue 9, 29, pp [8]. Jose M. Aller, Thoms G. Hbelter, Ronld G. Hrley, Sensorless speed mesurement of AC mchines using nlytic wvelet trnsform, IEEE Trnsctions on Industry Applictions, Vol. 38, Issue 5, 22, pp [9]. Tinyun Li, Yn Zho, Yongqing Hn, et l., Appliction of Hilbert-hung trnsform method in detection of hrmonic nd voltge flicer, Power System Technology, Vol. 29, Issue 2, 25, pp []. Tin-Yun Li, Yn Zho, Nn Li, A new method for power qulity detection bsed on HHT, Proceedings of the CSEE, Vol. 25, Issue 7, 25, pp []. D-Qin Ren, Shi-Xi Yng, Zho-Tong Wu, Instntneous frequency extrction method nd experiment bsed LMD, Journl of Zhejing University (Engineering Science), Vol. 43, Issue 3, 29, pp []. J. S. Smith, The locl men decomposition nd its ppliction to EEG perception dt, Journl of the Royl Society Interfce, Vol. 2, Issue 5, 25, pp [3]. Xio-Jun Zhu, Liu-Jun Fn, Shi-Qin Lv, Appliction reserch of LMD method in EEG signl processing, Computer Science, Vol. 39, Issue 2, 2, pp , 33. [4]. Jun-Sheng Cheng, Yu Yng, De-Jie Yu, The locl men decomposition method nd its ppliction to ger fult dignosis, Journl of Vibrtion Engineering, Vol. 22, Issue, 29, pp [5]. Wn-Ping Li, Peng Li, Cheng-Ji Liu, et l. HHTbsed hrmonic nd inter-hrmonic detection nd nlysis in microgrid, Shnxi Electric Power, Vol. 4, 2, pp Copyright, Interntionl Frequency Sensor Assocition (IFSA). All rights reserved. ( 389