I J E E E Interntionl Journl of Eletril, Eletronis ISSN No. (Online): 77-66 nd omputer Engineering (): 47-53(13) Investigtions on Power Qulity Disturnes Using Disrete Wvelet Trnsform hvn Jin, Shilendr Jin nd R.K. Nem Deprtment of Eletril Engineering, MNIT hopl, (M.P.) (Reeived 7 June, 13, epted 5ugest, 13) STRT: Extensive use of power eletroni devies nd non-liner lods in eletril power system use prolem of power qulity (PQ). Renewle energy soures re lso integrted to the grid through power eletronis sed equipment. So the power qulity issues re drwing ttention in reent yers. PQ disturne need to e deteted urtely It is lso essentil to find out the use of suh n event. Detetion nd lssifition of PQ disturne helps to ontrol suh event. In this pper, wvelet trnsform is pplied to notie, lolize, nd extrt power signl disturne. To study vrious power qulity disturnes, model simulted using MTL/Simulink toolox. The key pln underlying in the pproh is to deompose given disturne signl into lterntive signls tht represent smoothened version nd lose version of the first signl. Using multi resolution nlysis signl is deomposed. Keywords: Disrete Wvelet Trnsform (DWT), Momentry interruptions, Multiple resolution nlysis (MR), Voltge sg, Voltge swell, Wvelet Trnsform (WT). The si ide in wvelets is to nlyze signl I. INTRODUTION ording to sle rther thn frequeny. Wvelet hs een used for the nlysis of signl with disontinuities nd shrp spikes. y using wvelet multi resolution nlysis, proof is pitured y finite dd of elements t totlly different resolutions in order tht every element is dptively proessed supported the ojetive of the pplition [3-5]. This pper is divided in eight different setions. Definition nd disussion on power qulity disturnes re the min fous of setion II. nlytil methods used ommonly y different reserher to detet signl disturne re desried in setion III. Highlight on pplition res of wvelet presents in setion IV. In this model, vrious disturnes like voltge sg, swell, hrmonis nd interruption re simulted in setion V. Wvelet toolox is used to otin disrete wvelet trnsform on imported signl, otined through model results. Setion VII nd VIII re emphsizing on results nd onlusion respetively. The qulity of power supply hs eome mjor issue for eletril utilities nd eletriity onsumers. The poor qulity of the power supply my use mlfuntions of power servie equipments, instilities, short life time of equipments, nd so on. The power signl disturnes re lssified s impulse, nothes, glithes, momentry interruption, voltge sg, voltge swell, hrmoni distortion nd fliker. To improve the qulity of the power supply, it is required to detet soure of the disturnes urtely. The power qulity events should e deteted, lolized nd lssified urtely so tht proper mitigtion mesures ould e pplied [1, ]. Wvelet nlysis tehniques hve een implemented s new tool for fult detetion, loliztion nd lssifition of different power system trnsients [4, 7, 9, 1, 11]. ording to the literture, different wvelets n e used to deompose the signl. ommonly Duehies (D), iorthogonl (ior) nd oiflet hs used for identifying the imlne in tive power in power system. Wvelet fmilies n e found in detil in the referene [13], throwing light on properties of wvelets. Wvelets re mthemtil funtions tht divide dt into different frequeny omponents. These frequeny omponents re simple nd esy to study. II. POWER QULITY DISTURNES Detetion of power qulity disturnes hs eome mjor issue. ording to Interntionl Eletro tehnil ommission ( IE) impulsive nd osilltory trnsients, rief interruption, hrmoni distortion, voltge swell or sg re onsidered s disturnes.
The power qulity disturne is temporry devition in vlue from the stedy stte vlue due to sudden hnge of lod nd fults. Definition nd uses of different disturne re desrie elow ording to IEEE Std-1159, 15.. Voltge Sg Voltge sg is short-durtion derese of the Root Men Squre (RMS) voltge ( etween 1% to 9% ) tht lsts from.5 seonds to severl seonds. If it lsts for less thn hlf yle then it is onsidered s trnsient. Voltge sg results due to swithing opertions of lrge motors, lightning strokes nd trnsmission fults (disonnetion of supply). These momentry events n use omplete shutdown of power plnts, whih my tke hours to return to norml opertion [].. Voltge Swell n inrese in RMS vlue of voltge from 1.1p.u. to 1.8p.u., nd the lsting time is.5 period to 1min. is known s voltge swell. voltge swell our temporry, on the phse without fult of three phse iruit due to single line to ground fult. They n lso our on dding lrge pitor nk, removing lrge lod nd due to trnsfer of lods from one power soure to nother.. Hrmonis hrmoni is sinusoidl omponent of periodi wve or signl hving frequeny tht is n integer multiple of fundmentl frequeny. The term hrmoni refers to the deomposition of nonsinusoidl ut periodi signl into sum of sinusoidl omponents. Where n nd n re mplitude nd phse ngle for hrmoni order. Either time-domin or frequeny domin pproh n e used for hrmoni nlysis. Hrmonis re minly used y non-liner lods. Hrmoni distortion hs eome progressively more importnt in reent yers, due to the inrese in nonliner lods. D. Voltge Interruption Temporry interruptions my e thought of s voltge sgs with 1% mplitude or lrge perentge loss of voltge. Interruptions re used y trnsient fult. During short interruption, voltge level is to e Jin, Jin nd Nem 48 lose to zero. The uses my e lown fuse, or reker gp tht leds right down to the ending of the fility system set infliting lrge loss. E. Trnsients Voltge disturnes whih persist shorter thn sgs or swells re lssified s trnsients. These re used y the sudden hnges in the power system. The trnsient over voltge of durtion in the rnge of milliseonds re lled swithing surge nd in the rnge of miroseonds re lled impulse spike. III. NLYSIS METHODS. Fourier Trnsform (FT) The Fourier trnsform gives informtion out the presene of different frequeny omponents of the signl. It gives informtion of different frequeny omponent exists in the signl, ut FT do not provide informtion, when in time these frequeny omponents exist. For sttionry signl whose frequeny omponent does not hnge in time, this informtion is not required. Time frequeny representtion of signl is required for non sttionry signls. Short Time Fourier Trnsform ( STFT) is etter thn FT ut it gives fixed resolution ll the times due to fixed window size.. Wvelet Trnsform (WT) In wvelet nlysis, the use of totlly slle modulted window resolves the diffiulty of signlutting [5]. It gives vrile resolution. Trnsltion nd sling is used to generte wvelets from single si wvelet. This single si wvelet is lled s mother wvelet. In wvelet nlysis, we get detils nd pproximtions. The detils re the low sle, highfrequeny omponents nd the pproximtions re the high-sle, low-frequeny omponents of the signl. The strength of the wvelet nlysis is its ility of representing signls in ompt from nd in mny levels of resolutions. There re two types of wvelet trnsforms; disrete wvelet trnsforming ( DWT) nd ontinuous wvelet trnsform (WT). The prime differene etween disrete wvelet trnsform nd ontinuous wvelet trnsform is tht disrete wvelet trnsform use expliit suset of sle nd trnsltion vlues nd ontinuous wvelet trnsform uses ll possile sle nd trnsltion. In referene [9] uthor uses the ontinuous wvelet trnsforms to find out time durtion of the disturne nd the DWT to lulte the disturne mplitude.
. Disrete wvelet trnsforms (DWT) The signl is pssed through series of low pss nd high pss filter to generte disrete wvelet trnsform. Jin, Jin nd Nem 49 In filtering proess the originl signl is pssed through two omplementry filter nd we get two deomposed sequenes nd D. This proess produes DWT oeffiients. To extend the frequeny resolution, deomposition of signl is done repetedly nd signl n e relized into two lower frequeny rnges. This proess is known s multi resolution nlysis. pproximte oeffiients 1(n) nd detil oeffiients D1(n) otined fter pssing the signl through low pss nd high pss filters, t level 1. pproximtion oeffiients 1(n) re down smpled further with high nd low pss filters s shown in Fig.. The pproximte oeffiients (n) re then filtered gin to otin the next level of oeffiients. This filtering opertion progresses in this wy. Fig.. Signl three level deomposition Using MR. Nottion used in (), (3) nd (4) re y - originl signl, g high pss filter, h low pss filter. This deomposition hs hlved the time resolution sine only hlf of eh filter output hrterizes the signl. However, eh output hs hlf the frequeny nd of the input so the frequeny resolution hs een douled. t eh level in Fig. 3, the signl is deomposed into low nd high frequenies. Due to the deomposition proess the input signl must e multiple of n where n is the numer of levels. IV. WVELET PPLITION In mjor res of reserh suh s signl nd imge proessing, speeh disrimintions, dt ompression, Fig. 3. lok digrm of filter nlysis. de-noising nd numeril solution of differentil equtions, medil dignostis nd mny others wvelets hve een pplied suessfully [7]. In referene [1], use of disrete wvelet trnsform for evlution of the low frequeny eletromehnil osilltions in power systems nd found more ury in results during tive power imlne in power system.
Jin, Jin nd Nem 5 V. SIMULINK MODEL RETED TO STUDY VOLTGE SG, VOLTGE SWELL ND HRMONIS onn1 onn onn3 Susystem N n V I Three-Phse V-I Mesurement Three-Phse Trnsformer (Two Windings)1 Three-Phse Series RL Lod onn1 onn onn3 Susystem1 Three-Phse Trnsformer (Two Windings) ontinuous powergui Sope Sope Three-Phse Fult + v - Voltge Mesurement Sope1 simout To Workspe Fig. 4. simulink model. The disturnes like voltge sg, swell, hrmoni nd interruption re generted in simulink model shown in Fig. 4. Disrete wvelet trnsform n detet these disturnes nd provides informtion in oth time nd frequeny domin. Seletion of mother wvelet plys n importnt role in finding out nd lolizing the power qulity disturnes. s disussed in [1], for short nd fst trnsient disturnes, Du4 nd Du6 wvelets re good hoie, while for slow trnsient disturnes, Du8 nd Du1 re etter. Disrete wvelet trnsform is pplied nd four-level deomposition is done using duhies4. VI. WVELET RESULTS ording to IEEE stndrds, Duehies wvelet is very preise tool to monitor power qulity disturnes mong ll the wvelet fmilies. Signl otined fter using disrete wvelet trnsform, distint rippling rework for voltge sg, voltge swell nd hrmoni re unit displyed in Fig. 5, Fig. 6, Fig. 7 nd Fig. 8. Detil D1. -. 5 1 15 5 3 35 4 45 Detil D.5 -.5 5 1 15 5 Detil D3-4 6 8 1 1 Detil D4 5-5 1 3 4 5 6 7 pproximtion 4 1-1 1 3 4 5 6 7 Fig. 5. DWT of voltge sg.
The pproximtion revels the regulr pttern of the signl. Detils d1 is enough to produe preise time dt for ny type of disturne. Higher levels of deomposition re unit lrgely ustomed extrt dditionl dt needed for lssifition. From ll the 3 Figure, one n simply oserve strting nd ending time of ny disturne, the WTs hs terrily high vlue. Fig. 5 displying the pproximtion nd detiled version of voltge sg. From the Figure, sg n e Jin, Jin nd Nem 51 esily deteted t the lower sle. The strting time of sg is t 4 ms nd lsts till 4.5 ms nd therefore the durtion of sg event n e deteted nd lolized in lower levels. The detiled version of voltge swell in Fig..6 otined through finer resolution level. The voltge swell ours t 8 ms n e esily deteted s these disturnes perpetully found to possess pttern in initil level info. Detil D1-1 3 4 5 6 7 8 Detil D 5-5 5 1 15 5 3 35 4 Detil D3-5 1 15 Detil D4 1-1 4 6 8 1 1 pproximtion 4 1-1 4 6 8 1 1 Fig. 6. DWT of voltge swell. Detil D1-1 3 4 5 6 7 8 9 Detil D 5-5 5 1 15 5 3 35 4 45 Detil D3-5 1 15 5 Detil D4 1-1 4 6 8 1 1 pproximtion 4 1-1 4 6 8 1 1
The MR deomposition of voltge hrmonis re shown in Fig. 7. The detetion results t sle 1, showing disturne ourred t 1.34 ms nd end t 1.67ms. These disturnes lwys found to hve pttern in first level. Fig. 8 show the detiled version of four-level deomposition.voltge interruption is deteted nd lolized t first two finer resolution Fig. 7. DWT of voltge hrmonis. Jin, Jin nd Nem 5 levels. This is euse t the lower sles, the nlyzing wvelet is more lolized. The strting time of the interruption is.35 ms nd the ending time.45 ms n e oserved esily. The durtion of voltge interruption n e found out y differene of strting time nd ending time. Detil D1 5-5 1 3 4 5 6 Detil D 5-5 5 1 15 5 3 Detil D3 1-1 4 6 8 1 1 14 Detil D4-1 3 4 5 6 7 pproximtion 4 1-1 1 3 4 5 6 7 VII. ONLUSION The soures of power qulity disturnes must e known for improving the power qulity. The proposed tehnique found etter for detetion nd loliztion of power qulity disturnes. This tehnique urtely lotes voltge sg, swell, hrmoni distortion nd interruption. The results show orret lssifition of events will he performed y humn experts. Unlike Short Term Fourier Trnsform, the nlysis isn't depends on window size. This methodology will he extended to oserve other PQ events tht require to e monitored. This ould e utilized for hieving online, low ost PQ monitoring rel time system. Further rtifiil Neurl Network, Neuro fuzzy system, hyrid intelligent systems, spe vetor mhine et n e use for lssifition of disturnes y proessing the results otined through wvelet trnsform. Fig. 8. DWT of voltge Interruption. REFERENES [1]. Lin,.-H. Tso, M.-. "Power qulity detetion with lssifition enhnile wvelet-proilisti network in power system" Genertion, Trnsmission nd Distriution, IEEE Proeedings- 4 Nov. 5 Volume: 15,Issue: 6 On pges : 969-976 []. Dugn R.., MGrnghn M.F., Sntoso S. & ety H.W., Eletril Power Systems Qulity, MGrw Hill, ISN -7-1386-X, US, 3. [3]. Wvelet Toolox User s guide y The Mth works. [4]. Sntoso S., Powers E. J., Grdy W. M. nd Hoffmnn P., Power qulity ssessment vi wvelet trnsform nlysis, IEEE Trnstions on Power Delivery, Vol. 11, pp. 94 93, pr. 1996.
[5]. Relly Friendly Guide to Wvelets. Vlens, 1999.vlens@mindless.om [6]. Roi Polikr, 'The Wvelet Tutoril', 136 Rown Hll, Deprtment of Eletril nd omputer Engineering, Rown University, Glssoro, NJ 88, June 1996. [7]. M. Misiti, Y. Misiti, G. Oppenheim nd J.-M Poggi (6), Wvelet Toolox For Use With Mtl: user s guide version 3, Mthworks In. [8]. M.H.J. ollen, Understnding Power Qulity Prolems, Pistwy, NJ: IEEE Press. [9]. L. ngrisni, P. Dponte, M.D puzzo nd. Test, Mesurement Method sed on the Wvelet Trnsform for Power Qulity nlysis, IEEE Trns. Power Delivery, Vol. 13, pp.99-998, Ot. 1998. [1]. S. Sntoso, E.J. Powers, nd W.M. Grdy, Eletri power qulity disturne detetion Jin, Jin nd Nem 53 using wvelet trnsform nlysis, in Proeedings of the IEEE-SP Interntionl Symposzum on Tzme- Prequeny nd Time-Sle nlysis, Phildelphi, P, Ot. 5-8, 1994, pp. 166-169. [11]. D. Sxen, S.N. Singh nd K.S. Verm, Wvelet sed denoising of power qulity events for hrteriztion, Interntionl Journl of Engineering, Siene nd Tehnology, Vol. 3, No. 3, 11, pp. 119-13. [1]. Smir vdkovi, mir Nuhnovi, Mirz Kusljugi, Mustf Musi, Wvelet trnsform pplitions in power system dynmis, Eletri Power Systems Reserh, 83 (1) 37 45. [13]. Duehies, Ten Letures on Wvelets, Soiety for Industril nd pplied Mthemtis, Phildelphi, P, 199.