Optimal Power Dispatch of WECS and UPFC with ACO and ANFIS Algorithms

Transcription

1 Internatonal Journal on Electrcal Engneerng and Informatcs - Volume 10, Number 1, March 2018 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms Sunny Vg and Balwnder Sngh Surjan PEC Unversty of Technology, Chandgarh. sunnyv0417@gmal.coms Abstract: Wnd energy converson systems are the quckest developng renewable source of electrcal energy havng tremendous envronmental and socal advantages. In ths paper proposed an optmal reactve power dspatch (ORPD) technque of wnd power plants n grd connected power system. The ORPD ssue s a vtal ssue n the operaton of power frameworks. It s a nonlnear and mxed nteger programmng ssue, whch decdes deal qualtes for control parameters of reactve power makers to upgrade partcular target capactes whle fulfllng a few specalzed requrements. The ORPD technque conssts of ACO and ANFIS controller for mprovng the performance of multple wnd turbnes and the steady state stablty of the power system n terms of actve and reactve power flow of a transmsson lne. The ACO algorthm s used to extract the maxmum power from WECS. The generated maxmum power s transmtted through the UPFC connected HVDC lnk. The performance of the UPFC s mproved wth the ANFIS controller based on the actual and the reference power parameters of the grd connected power system. Furthermore, UPFC s utlzed to control actve and reactve flow of power n a transmsson lne. The proposed method s mplemented n MATLAB/Smulnk platform and tested wth the dfferent wnd speed condton. To verfy the effectveness of the proposed model, the obtaned results n the determnstc case s compared wth the exstng methods lke GA and PSO. Keywords: Unfed Power Flow Controller (UPFC), Ant Colony Optmzaton (ACO) Algorthm, Adaptve Neuro Fuzzy Inference System (ANFIS), wnd energy converson system (WECS) and power qualty (PQ). 1. Introducton The fast development of power system leads to ncrease n demand of electrcty whch leads to dfferent techncal problems lke PQ and stablty [1]. One of the most popular ways of generatng electrcty from renewable sources s to use wnd turbnes [2]. Wnd energy s effcent and promsng renewable energy resources n the world whch s contnuously growng wth the ncrease of electrcal power demand and the decrease n conventonal electrcty generaton resources [3, 4]. At the present tme and n the near future, generators for wnd turbnes wll synchronous generators, permanent magnet synchronous generators and nducton generators [5]. To mantan the operablty along wth relablty and power qualty of the grd, WECS are ntegrated wth the power grds rather operatng standalone. The grd code requrements are of two types: statc and dynamc requrements. Statc requrements nclude steady state behavor and power flow at the pont of common couplng (PCC) whereas dynamc grd code requrement ncludes the desred responses of the nducton generator durng grd dsturbances [6]. These requrements nclude the operatng range of voltage and frequency grd-support capablty, regulaton of power factor, and fault rde-through (FRT) capablty [7]. Accordng to the wnd turbne, Power qualty ssue s one of the major concerns n the dstrbuton network [8]. It s a measure of the standard of delvered power. By delverng low qualty electrc power to the consumer, t could affect the accuracy of utlty meterng; cause malfuncton to protectve relays; cause destructve damage to equpment and others. From Receved: September 21 st, Accepted: February 25 th, 2018 DOI: /jee

2 Sunny Vg, et al. research fndngs, power system montorng, nverter, dynamc voltage regulator (DVR), statc synchronous compensator (STATCOM), unfed power qualty condtoner (UPQC), and energy storage system are approaches that are used n allevatng power qualty ssues [9]. A UPFC s an electrcal devce for provdng fast-actng reactve power compensaton on hgh voltage electrcty transmsson networks [10]. UPFC provdes power flow control together wth ndependent voltage control. The man dsadvantage of ths devce s the hgh cost level due to the complex system setup [11]. By controllng the UPFC as a vrtual nductor, we am to ncrease the voltage at the termnals of the WECS and thereby mtgate the destablzng electrcal torque and power durng the fault [12]. The mpacts of wnd power on power qualty, the grd requrements for ntegraton of wnd turbnes, and dscussed the potental operaton and control methods to meet the challenges [13]. To mnmze the PQ ssues n wnd turbne framework, numerous sorts of power transformaton system to be assocated between the generator and grd lnes [14, 15]. The ORPD problem s a vtal ssue n the operaton of power frameworks. It s a nonlnear and mxed nteger programmng ssue, whch decdes deal qualtes for control parameters of reactve power makers to upgrade partcular target capactes whle fulfllng a few specalzed requrements [16, 17]. For solvng ths ssue, tradtonal methods are utlzed, such as, vares MPPT algorthm lke Hll clmbng search (HCS), ncremental conductance (INC) and optmum relaton based (ORB), perturb and dsturb (P&O) algorthm, soft computng based MPPT Fuzzy Logc Controller, Cuckoo Search Algorthm, Fuzzy logc [18, 19], Artfcal Neural Network and Artfcal Bee colony algorthm are adopted. Recently, the ntellgent search based optmzaton algorthms lke seeker optmzaton algorthm (SOA), Genetc Algorthm (GA), Partcle swarm optmzaton algorthm (PSO), harmony search algorthm, dfferental evolutonary based method, and gravtatonal search algorthm (GSA) have been created to manage the ORPD ssue [20]. In percepton of these ssues, ORPD ssue s a vtal ssue n the operaton of power frameworks. It s a nonlnear and mxed nteger programmng ssue, whch decdes deal qualtes for control parameters of reactve power makers to upgrade partcular target capactes whle fulfllng a few specalzed requrements. In the ORPD ssue, ntellgent search based optmzaton algorthms lke seeker optmzaton algorthm (SOA), harmony search algorthm, dfferental evolutonary based method, and GSA have been created to manage the ORPD ssue. However, there are a few specalzed ssues emergng from compensaton technques and the unpredctablty of control operaton. The recent research works are presented n Secton 2 and the detaled descrpton of the proposed technque s presented n Secton 3. The expermental results and dscusson are gven n Secton 4. Fnally, the Secton 5 concludes the paper. 2. Recent Research Works: A Bref Revew Numerous related works are already exsted n lterature whch based on Wnd energy converson system based grd for analyzng the power flow. Some of them revewed here. Joan Sau-Bassols et al. [21] have presented a seres tappng staton for ntegratng Offshore Wnd Power Plants (OWPP) nto a (Lne Commutated Converter Hgh Voltage Drect Current) LCC-HVDC transmsson system. The tappng staton allowed ntegratng the wnd power resources wthout buldng a HVDC lnk and t was based on a Current Source Converter (CSC). However, the CSC requres a mnmum DC current to extract the power comng from the OWPP whch may not be guaranteed dependng on the power condtons of the HVDC corrdor. In the reason, that paper proposed a coordnated operaton and control of the CSC and the OWPP. A steady-state analyss was performed to determne the approprate AC voltage level of the CSC. A power reducton algorthm was presented to lmt power extracton durng a reducton n the current of the HVDC transmsson system and under loss of communcatons between the CSC and the OWPP. 15

3 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms Fernando D. Banch et al. [22] have mplemented a coordnated control scheme n order that offshore WPPs connected through mult-termnal HVDC systems were able to contrbute to the prmary frequency control of the land AC grds. The proposed control scheme was capable of achevng a sutable frequency regulaton even under low wnd power condtons by allowng also the power share among AC areas. The control scheme was evaluated by dynamc smulatons n an adapted verson of the Cgr DC grd benchmark wth a fve-termnal HVDC grd ncludng two wnd farms and three AC networks. Chayan Bhattacharjee et al. [23] have developed an ntellgent extracton of optmum power and ts dspatch by usng fuzzy logc from a grd ted hybrd generaton system comprsng of a permanent magnet synchronous generator based wnd turbne and a low concentraton photovoltac generator. For photovoltac generator, maxmum power pont trackng control was mplemented usng fuzzy logc under varyng solar rradance. Power extracted from wnd turbne was desgned as a fuzzy functon of the dc lnk voltage error, ts rate of change and error n the drect axs current of the nverter. That reduces hgh frequency oscllatons n the wnd extracted power. A falure mode and effect analyss was done for power converters and possble mtgaton schemes were suggested for dfferent faults. Ast Mohanty et al. [24] have presented a comparatve study of transent stablty and reactve power compensaton ssues n an autonomous wnd-desel-photovoltac based hybrd system (HS) usng robust fuzzy-sldng mode based UPFC. A lnearze small-sgnal model of the dfferent elements of the HS was consdered for the transent stablty analyss n the HS under varyng loadng condtons. An IEEE type 1 exctaton system was consdered for the synchronous generator n the HS, wth neglgble saturaton characterstc, for detaled voltage stablty analyss. It was noted from the smulaton results that the performance of UPFC was superor to statc VAR compensator and statc synchronous compensator n mprovng the voltage profle of the HS. Further, fuzzy and fuzzy-sldng mode based UPFC controller was desgned n order to mprove the transent performance. Rong Zeng et al. [25] have developed the control and operaton of a hybrd HVDC system comprsng a wnd farm-sde voltage-source converter rectfer and a grd-sde LCC nverter for ntegratng wnd power. The confguraton and operaton prncple of the hybrd HVDC system were descrbed. Commutaton falure n the LCC nverter durng an ac network dsturbance was consdered and ts mpact on the hybrd system operaton was analyzed. An enhanced control strategy for the LCC nverter at the grd sde and an alternatve modular multlevel converter (MMC) topology usng mxed half-brdge and full-brdge modules consdered for the rectfer at the wnd farm sde were proposed. Recently, wnd vtalty has gotten to be a standout amongst the most essental and promsng wellsprngs of renewable vtalty, whch requests extra transmsson lmt and better method for keepng up framework relablty. The need to coordnate the renewable energy lke wnd energy nto power system s to make t concevable to mnmze the ecologcal effects. The coordnaton of wnd energy nto exstng power system ntroduces a specalzed dffcultes and that requres thought of voltage regulaton, stablty, power qualty problems lke voltage sag, voltage swell, nose and harmoncs. The produced power PQ s sgnfcantly nfluenced by operaton of a transmsson and dstrbuton system. To enhance the power, effectveness of grd connected WECS; hgh profcency gadgets takng nto account power hardware types of gear have been progressvely utlzed, for example, STATCOM, UPQC and so on. In the paper, UPFC s utlzed to control actve and reactve flow of power n a transmsson lne. It s a mx of arrangement assocated Statc Synchronous Seres Compensator (SSSC) and shunt connected Statc synchronous Compensator (STATCOM). Here, the ACO Algorthm, and ANFIS approach s employed to decde the fnest compromse soluton. The ACO algorthm s used to extract maxmum wnd power and ANFIS s utlzed to enhance performance of UPFC for accomplshng the optmal results. The detaled descrpton of the proposed method s descrbed n the secton 3. 16

4 Sunny Vg, et al. 3. Proposed Methodology for Multple Wnd Turbne Wth UPFC In ths secton dscussed about the performance analyss of the proposed methodology of the wnd turbne wth UPFC. The proposed methodology conssts of ACO and ANFIS algorthm for the steady state stablty of the power system and mprovng the performance of multple wnd turbnes. The ACO Algorthm s used as an optmzng technque to extract maxmum wnd power by controllng ptch angle and UPFC n Co-ordnaton wth HVDC lnk s used to control actve and reactve flow of power n a transmsson lne. The proposed controllng model s llustrated n fgure 1, whch conssts of proposed control system and grd connected power system. The wnd energy converson system has four major components bult wth a wnd turbne, generator, nterconnecton apparatus and control systems. The wnd turbne operates ether down-wnd or up-wnd. Most turbne manufacturers have opted for reducton gears between the low speed turbne rotor and the hgh speed three-phase generators. To generate the maxmum power the ACO algorthm s used to extract the maxmum power nstead of MPPT algorthm. In ths proposed algorthm the ptch angle s controlled and extracts the maxmum power. The ptch angle s kept constant at zero degree untl the speed reaches maxmum speed of the trackng characterstc. The ptch angle s proportonal to the speed devaton of the wnd turbne speed. Ptch angle Wnd Energy System HVDC lnk Grd connected Power system Sref Proposed ACO Algorthm Sact P ref UPFC Proposed ANFIS Controller Control pulse P act Fgure 1. The proposed model for optmal power dspatch from WECS Then the power s transmtted to grd connected power system through HVDC lnk. The Optmzaton of electrc power transmsson system capacty requres a relable power flow controller. The power flow controllers must be able to control the level of electrcal voltage and actve and reactve power flow wthout reducng the level of stablty and securty of the transmsson system. Latest technology n the control of power flow s a UPFC, whch s mantanng the steady state stablty of the system. UPFC conssts of a combnaton of a shunt and seres branches connected through the DC capactor The seres connected nverter njects a voltage wth controllable magntude and phase angle n seres wth the transmsson lne, therefore provdng real and reactve power to the transmsson lne. The shunt-connected nverter provdes the real power drawn by the seres branch and the losses and can ndependently provde reactve compensaton to the system. For ncreasng the performance of the UPFC the ANFIS algorthm s used to control the power varaton of the grd connected power system. In ths algorthm compensate the varatons between the actual power and reference power of the power system. The modelng of the proposed model s explaned n a followng secton. 17

5 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms A. Mathematcal Modelng of Wnd Turbne The wnd turbne s dstngushed by no dmensonal curves of the power coeffcent C as a utlty of together the tp speed rato and the blade ptch angle. So as to completely explot the exstng wnd energy, tp speed rato should be preserved at ts fnest value. Consequently, the power coeffcent equvalent to that value wll turn out to be hgh. It can be dstnct as the proporton of the angular rotor speed of the wnd turbne to the lnear wnd speed at the tp of the blades [26]. It can be conveyed as equaton (1), t R (1) V In addton, the relaton between and can be found n the followng equaton (2), (2) A generc equaton s used to model C (, ). The characterstcs of the wnd turbne model s gven as equaton (3), C p 5 C 2 C p (, ) C1 C3 C4 e C6 (3) The mechancal power of the WT s a utlty of producer speed for dssmlar wnd speeds and for blade ptch angle 0. The automatc power detaned by a wnd turbne n the fxed condton s specfed by equaton (4), A 3 Pm C p (, ) V (4) wnd 2 From the equaton(4), Pm s the automatc output power of the turbne (W), C p s the presentaton coeffcent of the turbne, s the ar densty (kg/m3), A s the turbne swept area (m2), V wnd s the wnd speed (m/s), s the tp speed proporton and s the blade ptch angle ( ). Snce on top of the equaton can be standardzed and n the per unt ( pu ) scheme we have specfed as equaton (5), 3 P k C V (5) Where, m _ pu p p _ pu wnd _ pu P m _ pu C p _ pu s the presentaton coeffcent n pu depend on the hghest value of p and A, s the power n pu derved from the ostensble power for exactng values of V wnd _ pus the wnd speed n pu depend on the base wnd speed, p fewer than or dentcal to 1 for C p _ pu 1pu andvwnd _ pu 1pu. p C. k s the power gan whch s B. Maxmzaton of Wnd Power Generaton The adaptaton of wnd energy s offered by the wnd turbne and the nducton generator (WTIG). The stator wndng s assocated openly to the network and the motor s determned by the wnd turbne. The power detaned by the wnd turbne s transformed nto electrcal power by the nducton generator and s conveyed to the network by the stator wndng. The ptch angle s prohbted to control the producer output power to ts nsgnfcant value for elevated wnd speeds. To produce power the nducton producer speed must be vaguely hgher than the synchronous speed. Excludng the speed dssmlarty s usually so dmnutve that the WTIG s measured to be a preset speed wnd generator. The mmedate power mmersed by the nducton generator can be suppled by the network [27]. At ths tme, an UPFC s derved from 18

6 Sunny Vg, et al. the current-source converter topology s projected, whch permts an optmzed mmedate power nserton for the duraton of voltage dps take place n the network, and assst the preset speed WTIG to produce ts mortal voltage. To adjust the speed derved from the ptch angle of the power scheme s depcted n subsequent segment. Max Wnd Turbne Electrc power Power transducer Measured power Proposed Ptch angle controller 0 0 Power set-pont Blade ptch Proposed Ptch angle controller Ptch demand Fgure 2. The control loop for regulatng speed and ptch angle n WT The foremost fundamentals of the control loop for a preset speed ptch synchronzed turbne are exposed n fgure 2. Generally, the Proportonal Integral (PI) controller s employed to manage the blade ptch angle to mantan the electrc output power to the ostensble automatc power. The ptch angle s reserved stable at zero when the consdered electrc output power s below ts ostensble value. The ostensble value s assessed derved from the ACO algorthm, whch manage the fnest explanaton n non-lnear envronment. When t augments hgher than ts ostensble value the regulator augment the ptch angle to renstate the consdered power to ts ostensble value [28]. Afterward the WT representaton s clarfed n a subsequent segment. C. Proposed ACO algorthm for maxmum power generaton For optmzng wnd power usng ACO Algorthm dfferent parameters s consdered such as rotor angle, tp speed rato, wnd speed and ar densty. By takng nto consderaton dfferent parameters ACO as decson makng tool optmal value of wnd speed where fnd out the maxmum power from the wnd turbne. Then, the ftness functon s evaluated to the nput parameters. Here, the maxmum value of power s taken as a ftness functon s gven as equaton (6), Ft max( P m ) (6) Where, P ) Ft max( m s the ftness functon of ACO algorthm, m P s the maxmum power from the wnd turbne. The ACO makes use of agents, called ants, whch mmc the behavor of real ants n how they manage to establsh shortest- route paths from ther colony to feedng sources and back. Ants communcate nformaton through pheromone trals, whch nfluence whch routes the ants follow, and eventually lead to a soluton route [29]. Intally, The ACO was desgned to solve the Travelng Salesman Problem (TSP). In the TSP, a gven set of n ctes has to be vsted exactly once and the tour ends n the ntal cty. We call d j (, j 1,2,..., n) the length of the path between ctes and j. In the case of Eucldean TSP, d j dj s the Eucldean dstance between and j s gven as equaton (7), j 2 x x (7) 19

7 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms The ctes and routes between them can be represented as a connected graph ( n, E), where n the set of towns and E s the set of edges between towns (a fully connected graph n the Eucldean TSP) [30]. The ants move from one cty to another followng the pheromone trals on the edges. Let (t) be the tral ntensty on edge (, j) at teratont. Then, each j ant k( k 1,2,..., m) chooses the next cty to vst dependng on the ntensty of the assocated tral. When the ants have completed ther cty tours, the tral ntensty s updated accordng to equaton (8), j ( t 1). j ( t) j, t 1,2,..., T (8) Where, s a coeffcent such that ( 1 ) represents the evaporaton of tral between teraton t and t 1, T s the total s the number of teratons (generaton cycles) the tral ntensty s gven as equaton (9), m k (9) j k1 k j Where, s the quantty per unt of length of tral substance (pheromone n real ants) lad on edge (, j) by the go to wth a probablty (t) th k ant between teraton t and t 1 p k j.an ant k at cty chooses the cty j to, whch s a functon of the town dstance and of the amount of pheromone tral present on the connectng edge. The probablty functon s gven as equaton (10), j ( t). j k j U k pj ( t) k ( t). k ku otherwse 0 (10) Where, U s a set of the ctes can be chosen by the ant at cty for the next step, j 1 d j k s a heurstc functon whch s defned as the vsblty of the path between ctes and j ; parameters and determne the relatve nfluence of the tral nformaton and the vsblty [31]. The flow dagram of the proposed ACO algorthm s presented n fgure 3. Start th k Parameter ntalzaton Construct a soluton for each ant Update pheromone values Select feature subset Update best ant wth ftness value No Is Condton satsfed Yes Select best ant s features End Fgure 3. The flow dagram of the proposed ACO algorthm 20

8 Sunny Vg, et al. Steps for the Proposed ACO Algorthm Step 1: Intaton. The amount of the pheromone on each sde s ntated nto a tny constant value; allocate m ants randomly to n ctes. Step 2: In ACO, the pseudorandom proportonal rule s used, the probablty for an ant to move from cty to cty j depends on a random varable q unformly dstrbuted over [0, 1], and a predefned parameter q 0. However, the random value s descrbed as equaton (11), max arg fq q j ( ) j, j 0 j (11) J otherwse Here, j s a random varable determned n accordance wth above equatons. Ths strategy obvously ncreases the varety of any searchng, thus avodng any premature fallng nto the local optmal soluton and gettng bogged down. Step 3: The local pheromone update s performed by all the ants after each constructon step. Each ant apples t only to the chosen cty s gven as equaton (12), ( t 1) (1 ). ( t) (12) j j 0 Where, 0 1s a decay parameter, 1 0 s the ntal values of the pheromone n..l nn trals, n s the number of ctes n the TSP and L s the cost produced by the nearest neghbor heurstc. The probablty equaton s manly to avod very strong pheromone paths to be chosen by other ants and to ncrease the exploratve probablty for other paths. Once the edge between cty and cty j has been vsted by all ants, the local updatng rule makes pheromone level dmnsh on the edge. So, the effect of the local updatng rule s to make an already edge less desrable for a followng ant. Step 4: Computng of the optmal path. After m ants have travelled through all the ctes, compute the length of the optmal. Step 5: Global updatng of pheromone. After all the ants have travelled through all the ctes, update only the amount of the pheromone on the optmal path wth equaton (13) and (14), ( t 1) (1 ). ( t). ( t) (13) j j j 1 best f (, j) G t L j ( ) (14) gb 0 otherwse Where, s constant and L s the length of global best tour. gb Step 6: If the desgnated search number s not attaned, then repeat the above steps. To vary the parameters are based on our requrements and get the mproved performance of the algorthm. From the proposed algorthm the ptch angle s controlled and the maxmum power s reached wth the maxmum teraton. Then the power s transmtted to the grd connecter power system through HVDC lnk. D. HVDC Functonaltes for Power Transmsson Durng the state of power exchange n nterconnected lnes to a substaton under varable or constant power, the HVDC converters comprehends the power converson and later stablzes the voltage through the lnes gvng a break even margn n the power transmsson. The operaton of HVDC flters any system harmoncs developed n the network and mproves the nn 21

9 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms power transmsson to the recevng end by ndependently adjustng the real and reactve power control. The sgnfcance of HVDC controller consdered as part of UPFC devce s a structure of the back-to-back converter that governs the converson of ac-dc-ac. HVDC s assgned for frequency and phase ndependent short or long dstance overhead or underground bulk power transmsson wth hgh speed controllablty [32]. Ths provdes greater real power transmsson and less mantenance. It reduces the chances of nstallng power cables especally n dffcult transmsson that travels under water. By makng use of the back-to-back converters, power transmsson under non-synchronous ac systems s easly adaptable. The nstallaton of HVDC also depends on the dc voltage and current ratngs desred n the network that Yelds for optmum converter cost. The DC overhead lnes or cables are lnked to AC buses and network. The operaton of HVDC s restrcted when network system contans low short crcut ratos. Therefore, nsulaton n the HVDC s essental n such cases. However, ths does not restrct the converter statons operaton [33]. The HVDC nsulaton must wthstand the stress produced n ac and dc voltages to allow full operaton of HVDC n the lnes. E. Power Flow Analyss of UPFC Model Generally, UPFC s the best effect on effcent steady state transmsson. Because of ts desgn and workng prncple t s havng such technology. Ths technology s settlng effect on steady state, dynamc and transent stabltes. The major components of UPFC are two AC/DC converters, seres and shunt transformer and the capactor. One AC/DC converter s connected n seres along the transmsson lne over a seres. And the other s connected parallel wth the transmsson lne through shunt transformer. The DC output sde of the both converters s connected wth the capactor. Ths capactor gves DC voltage for the converter operaton [34]. The two voltage source converters of the UPFC, connected through a D.C lnk can be modeled as two deal voltage sources, one connected n seres and the other n shunt between the two buses. The output of the seres voltage sourcev and are controllable magntude and angle between the lmtsv max se se mn se se se V V and 0 2 respectvely and of the shunt voltage se max mn source svsh and sh controllable between the lmtsvsh Vsh V and sh 0 sh 2. Z se and Z are the mpedances of the two couplng transformer one connected n seres and other sh n shunt between the lne and the UPFC. The equvalent crcut of ths model s depcted n fgure 4. The seres and parallel voltages are gven by equaton (15) and (16), V V (cos( ) jsn( )) (15) V se sh se se se V (cos( ) jsn( )) (16) sh sh sh mn max Where, V ( V V V ) and ( 0 2 ) are voltage magntude and adjustable cr cr cr cr cr cr angle of seres voltage resource. The voltage magntude and equvalent angle ( vr vr v, ) desgnate the equvalent voltage source lmtaton [35]. Pk jq k Z se Pse jq se P jq m m I k I se I m bus k Psh jq sh Z sh Pse P sh bus m V m V k V sh Fgure 4. The equvalent dagram of UPFC 22

10 Sunny Vg, et al. The nput system data ncludes the basc system data needed for conventonal power flow calculaton consstng of the number and types of buses, transmsson lne data, generaton and load data, locaton of UPFC and the control varables of UPFC.e the magntude and angles of voltage output V and V of two converters. The ncluson of the UPFC ncreases one bus n se sh the system. The UPFC power equatons are combned wth the network equatons to gve equaton (17), P jq Where, P n j1 V, V Y ( ) P jq (17) j j j j jq s actve and reactve power flow due to UPFC between the bus k and m. P jq denotes actve and reactve power at the th bus. V refers voltage and angle th th of bus, V j refers Voltage and angle of j bus. Y j j s a admttance of the transmsson lne between the bus and j. The power flow constrant of the UPFC s ncluded n the jacoban. The ncluson of these varables ncreases the dmenson of the jacoban. The power equatons are msmatched untl convergence s acheved. A scalar multpler s used to control the updatng of varables to ensure that they converge n an optmal way to the soluton pont. The real and reactve powers of the partcular bus are descrbed n the followng equatons (18) and (19). P V V Q V V Where, of buses, NB j n1 NB V and j j n1 G j j cos B sn (18) j j j j j G sn B cos s (19) j j V symbolze the voltage of and j buses respectvely, NB s the total number s the angle between and j buses respectvely, G and B s the conductance and susceptance values respectvely. At ths moment the complete scheme of UPFC s profcent of mutually delverng and captvatng actual and mmedate power from the scheme. The power stablty among the seres and shunt converter s a prerequste to mantan a constant voltage transversely the DC capactor. The power flow capablty and temporary constancy are enhanced by seres dvson of UPFC whch nsert the voltage of varable magntude and phase angle. These seres dvson can replace actual power among a broadcast lne to develop above sad emnence of broadcast lne. The shunt dvson of UPFC scheme s replaces a current and power factor angle of convenent magntude through the power scheme [36]. It s normally prohbted to stablty the actual power combnaton or nserton nto the power scheme by the seres dvson, n addton the wounded by changeable the DC voltage at the preferred value. F. UPFC Enhancement wth ad of ANFIS ANFIS s an adaptve network that s generally correspondng to a fuzzy presumpton scheme, where the productvty has been attaned by utlzng fuzzy regulatons on nputs. An ANFIS scheme s a mxture of neural network and fuzzy scheme n such a manner that neural network s utlzed to resolve the lmtaton of fuzzy scheme. A neural network s employed to mechancally change the scheme lmtaton. The ANFIS s a very domnant method for formng nonlnear and compound scheme through less nput and output preparaton data through qucker learnng and hgh exactness. The neuro-fuzzy scheme through the learnng competence of neural network and by the compensaton of the rule-base fuzzy scheme can develop the presentaton consderably and can supply a system to ntegrate precedent explanaton nto the categorzaton procedure [37]. In neural network the preparaton fundamentally constructs the j j 23

11 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms scheme. Though, utlzng a neuro fuzzy format, the scheme s constructed by fuzzy logc descrpton and s after that dstngushed by neural network preparaton algorthms. 1. Archtecture of ANFIS The representaton method utlzed by ANFIS s comparable to numerous scheme recognton procedures. Intally, a parameterzed representaton arrangement (connectng nputs to assocaton utlty to regulated outputs to assocaton task, and so on) s assumed. After that nput/output data s composed n a structure that wll be functonal by ANFIS for preparaton. ANFIS can subsequently be utlzed to gude the FIS representaton to follow the preparaton data obtanable to t by alterng the assocaton task lmtaton along wth an elected mstake prncple. Functon of ANFIS looks lke feed-forward back propagaton network. Resultant lmtatons are consdered onward whereas bass lmtatons are ntended backward. There are two learnng process n neural segment of the scheme: Hybrd learnng process and back-propagaton learnng process. In fuzzy segment, only zero or prmary order Sugeno nference scheme or Tsukamoto presumpton scheme can be utlzed. Ths segment ntates the fundamentals of ANFIS network desgn and ts hybrd learnng regulaton [38]. The Sugeno fuzzy model was proposed by Takag, Sugeno, and Kang n an effort to formalze a systematc approach to generatng fuzzy rules from an nput output dataset. The propose technque objectve functon s evaluated by equaton (20), ref act F Mn( E) Mn{ P P } (20) ref act Where, E s the error value between the reference and actual powers P and P represent the reference and actual values of power. The ANFIS structure of two nput one output s depcted n fgure 5. The two nputs are error E, change n error E and power varaton output. The change n error s estmated as equaton (21), E E( s) E( s 1) (21) Where, E( s 1) s the prevous state of error, In ths connected structure, the nput and output nodes represent the tranng values and the predcted values, respectvely, and n the hdden layers, there are nodes functonng as membershp functons (MFs) and rules. Ths archtecture has the beneft that t elmnates the dsadvantage of a normal feed forward multlayer network, where t s dffcult for an observer to understand or modfy the network. Here E, E are nputs, Y s output, the crcles represent fxed node functons and squares represent adaptve node functons. Consder a frst order Sugeno-fuzzy nference system whch contans two rules: Rule 1: If E s A1 and E s B 1; then f1 p1e q1e r1 Rule 2: If E s A2 and E s B 2 ; then f 2 p2e q2e r2 Layer 1 Layer 2 Layer 3 Layer 4 Layer 5 A 1 E Error n W1 W1 N X1 X 2 A n W 1 f 1 N Y Output B 1 E W2 n Change of error N W 2 X1 X 2 W 1f 2 B n Fgure 5. The controllng structure of ANFIS 24

12 Sunny Vg, et al. At ths pont, p1, p2, q1, q2, r1 and r2 are lnear lmtaton and A1, A2, B1 and B2 are nonlnear lmtaton. ANFIS s an executon of a fuzzy logc presumpton scheme through the structural desgn of a fve-layer feed-forward network. The scheme structural desgn contans fve layers,.e., fuzzy layer, product layer, normalzed layer, de-fuzzy layer and total output layer. By ths manner ANFIS utlze the compensaton of learnng competence of neural networks and presumpton system related to human bran offered by fuzzy logc. The projected ANFIS structural desgn contans fve layers n whch crcle shaped nodes are descrbed as fxed nodes, whch means the node lmtaton are autonomous on the further nodes and square shaped nodes are descrbed as adaptve nodes, whose node lmtaton rely on the supplementary nodes [39]. Layer 1: Quantty of nodes n ths level symbolzes the quantty of fuzzy sets utlzed. The frst layer nodes are n square shaped and lmtatons of these nodes sgnfcant the assocaton task. In ths study work, a trangular assocaton task s utlzed as equaton (22), O E); O ( E); 1,2,...,5 (22) 1 A ( 2 B Here, E and E node. O 1 and are the two nputs, A and B are the fuzzy sets correspondng to the O 2 are membershp functons of fuzzy set A and B. Generalzed bell membershp functon s popular method for specfyng fuzzy sets because of ther smoothness and concse notaton s defned as equaton (23), 1 (23) A ( E) 2b E c 1 a Now, a, b and c s the lmtaton group of the assocaton task. The center and wdth of the assocaton task s dverse by regulatng slopes at the ntersect ponts. c anda. The lmtaton b s utlzed to drect the Layer 2: In ths level, nodes are recognzed as regulaton nodes. Each node ncreases the nward sgnals and offer the product that sgnfy the weght (w) of a regulaton and s specfed by equaton (24), w ( E) ( E); 1,2,...,5 (24) A B Layer 3: In ths layer, weghts are normalzed as equaton (25), w w ; 1,2 (25) w w1 Layer 4: Generally, the measurement of ths level resolves the nteger of fuzzy regulatons utlzed n the scheme. For each node of ths level, output(y) s specfed by equaton (26), Y w f (26) Where, f p E q E r and w s the standardzed weght of th node. p, q and r are the ntended lmtaton that are resolute throughout the preparaton procedure and s the quantty of assocaton task of partcular nput. Layer 5: In the fnal layer, the overall output s computed by addng all the outputs of the prevous layer. The sngle fxed node calculates the overall output as the summaton of all ncomng sgnals as equaton (27), 25

13 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms w f w f (27) w It can be expermental that there are two adaptve levels n ths ANFIS structural desgn,.e. the frst level and the fourth level [40]. The projected regulator s verfed n MATLAB/ANFIS edtor tool box among a trangular assocaton task as t suggests smallest preparaton fault. Because, the back propagaton algorthm s dsreputable for ts lateness and nclnaton to develop nto fascnated n lmted mnma, a fuson learnng algorthm s utlzed n ths donaton. Ths algorthm s quck and precse n recognze the lmtaton. 4. Results and Dscussons In ths secton, we dscussed about the performance analyss of the proposed technque and compared wth some dfferent technques lke GA and PSO. The proposed technque s appled wth Intel(R) core(tm) 5 processor, 4GB RAM and MATLAB/Smulnk (R2015a) platform. The Smulnk model of the proposed system s llustrated n the fgure 6, whch shows the small sgnal stablty model wth the MG connected load system s controlled based on the proposed technque. The proposed ACO algorthm s used to gather the maxmum power from the WECS and ANFIS s used to control the power flow of the grd connected power system wth the help of UPFC. Fgure 6. The Smulnk model of the proposed system 26

14 Sunny Vg, et al. () () () 27

15 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms (v) Fgure 7. Performance analyss of wnd energy converson parameters () wnd speed, () wnd turbne rotor speed, () ptch angle, (v) HVDC lnk voltage (a) (b) 28

16 Sunny Vg, et al. (c) (d) Fgure 8. Power performance of (a) actve power, (b) reactve power of wnd and (c) actve power, (d) reactve power of UPFC model The performance of the WECS generated power s based on the speed of the wnd turbne, so to control the wnd turbne speed only to get the adequate power from the WECS. Based on the wnd speed the WECS s generated the power, to mprove the PQ to controls the ptch angle of the wnd turbne rotor for stable power generaton. The measured wnd speed, turbne speed, ptch angle and the output HVDC lnk voltage are llustrated n fgure 7. Then the generated power transmtted to grd usng the HVDC lnk and UPFC. The actve and reactve power of the WECS and UPFC s descrbed n a fgure 8. 29

17 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms () () () 30

18 Sunny Vg, et al. (v) Fgure 9. The grd power performance wth () actve power () reactve power, () postve sequence and (v) negatve sequence of grd connected wnd power system (a) (b) 31

19 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms (c) (d) (e) Fgure 10. Comparson analyss of grd (a) actve power, (b) reactve power, (c) actve power, (d) reactve power of the UPFC and (d) the HVDC lnk voltage The wnd turbne model s generates the actve power based on the wnd speed and the rotor angle of the turbne. To control the ptch angle of the wnd turbne s gatherng the maxmum and lnear power from the power system. The generated power s transmtted to grd based on the HVDC lnk. The actve and reactve power of the grd connected proposed system s fgure 9. Whch s also stablzng the output power for gathered the stablzed power. 32

20 Sunny Vg, et al. Fnally, to prove the effcency of the proposed technque s compared wth some exstng technques lke GA and PSO. The comparsons of the technques are performed n a grd sde and UPFC. To measure the actve and reactve power of the grd and UPFC and the HVDC lnk voltage s descrbed n fgure 10. The comparson s based on the grd actve and reactve power and UPFC real and reactve power of the proposed system. Then the HVDC lnk voltage s reached the stable poston wthn 0.05sec. From the above the observed results, the proposed ACO based WECS and ANFIS based UPQC controller acheves better performance for compensatng the PQ. The dc bus voltages are almost mantaned to the reference value under all dsturbances. To mprove the performance of the proposed system s to stablze the power flow of the grd connected system. 5. Concluson In ths paper presented an optmal reactve power dspatch (ORPD) technque wth WECS and UPFC for requred power delver to the grd connected power system. In ths proposed technque utlzed the ACO algorthm and ANFIS controller for mprovng the steady state stablty of the power system. The ACO algorthm s utlzed for gatherng maxmum power from the WECS. Based on the algorthm s control the ptch angle of the wnd turbne model for extractng maxmum power from the WECS. The maxmum generated power s transmtted through HVDA lnk to grd connected power system. To optmzng the output power of the system s based on the UPFC, whch s control actve and reactve flow of power n a HVDC transmsson lne. The ANFIS controller s employed to enhance the performance of the UPFC. The UPFC optmzed the varatons of the actual power and reference power of the grd connected power system. The proposed optmzaton problem s solved utlzng ACO and ANFIS approach, whch s employed to decde the fnest compromse soluton. The proposed method s mplemented n MATLAB/Smulnk platform and tested wth the dfferent wnd speed condton. To verfy the effectveness and optmalty of the proposed model, the obtaned results n the determnstc case s compared wth the exstng methods lke GA and PSO. 6. Reference [1]. S.G. Bharath Dasan, Sharon Ravchandran Kamesh and R.P. Kumudn Dev, "Steadystate analyss of Grd connected WECS usng FACTS controller", In proceedngs of IEEE conference on Emergng Trends n Electrcal and Computer Technology, pp , 2011 [2]. Dpesh Kumar and Kalyan Chatterjee, "A revew of conventonal and advanced MPPT algorthms for wnd energy systems", An Internatonal Journal of Renewable and Sustanable Energy Revews, Vol. 55, pp , 2016 [3]. Yasser M. Alharb, A. M. Shddq Yunus and A. Abu Sada, "Applcaton of UPFC to Improve the FRT Capablty of Wnd Turbne Generator ", An Internatonal Journal of Electrcal Energy, Vol. 1, No. 4, pp , 2013 [4]. Yazhou Le, Alan Mullane, Gordon Lghtbody and Robert Yacamn, "Modelng of the wnd turbne wth a doubly fed nducton generator for grd ntegraton studes", IEEE transactons on energy converson, Vol. 21, No. 1, pp , 2006 [5]. Beltran, Brce, Tarek Ahmed-Al and Mohamed El Hachem Benbouzd, "Sldng mode power control of varable-speed wnd energy converson systems", IEEE Transactons on Energy Converson, Vol. 23, No. 2, pp , 2008 [6]. Frede Blaabjerg, Remus Teodorescu, Marco Lserre and Adran V. Tmbus, "Overvew of control and grd synchronzaton for dstrbuted power generaton systems", IEEE Transactons on ndustral electroncs, Vol. 53, No. 5, pp , 2006 [7]. Snehaprava Swan and Pravat Kumar Ray, "Short crcut fault analyss n a grd connected DFIG based wnd energy system wth actve crowbar protecton crcut for rde through 33

21 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms capablty and power qualty mprovement", An Internatonal Journal of Electrcal Power and Energy Systems, Vol. 84, pp , 2017 [8]. GM Joseln Herbert, Selvaraj Inyan, E. Sreevalsan and S. Rajapandan, "A revew of wnd energy technologes", An Internatonal Journal of Renewable and sustanable energy Revews, Vol. 11, No. 6, pp , 2007 [9]. Ken Weng Kow, Yee Wan Wong, Rajparthban Kumar Rajkumar and Rajprasad Kumar Rajkumar, "A revew on performance of artfcal ntellgence and conventonal method n mtgatng PV grd-ted related power qualty events", An Internatonal Journal of Renewable and Sustanable Energy Revews, Vol. 56, pp , 2016 [10]. M. Ramesh and A. Jaya Laxm, "Enhancement of Power Transmsson Capablty of HVDC System usng Facts Controllers", An Internatonal Journal of Advances n Engneerng and Technology, Vol. 1, No. 5, pp , 2011 [11]. N. Senthl kumar and J. Gokulakrshnan, "Impact of FACTS controllers on the stablty of power systems connected wth doubly fed nducton generators", An Internatonal Journal of Electrcal Power and Energy Systems, Vol. 33, No. 5, , 2011 [12]. Chen, Zhe, "Issues of connectng wnd farms nto power systems", In proceedngs of IEEE conference on Transmsson and Dstrbuton Conference and Exposton, pp. 1-6, 2005 [13]. Marco Dorgo and Thomas Stützle, "The ant colony optmzaton metaheurstc: Algorthms, applcatons, and advances", Handbook of metaheurstcs, pp , 2003 [14]. Goms-Bellmunt, Orol, Adra Junyent-Ferre, Andreas Sumper and Joan Bergas-Jane, "Control of a wnd farm based on synchronous generators wth a central HVDC-VSC converter", IEEE Transactons on power systems, Vol. 26, No. 3, pp , 2011 [15]. Om Prakash Mahelan and Abdul Gafoor Shak, "Comprehensve overvew of grd nterfaced wnd energygeneraton systems", An Internatonal Journal of Renewable and Sustanable Energy Revews, Vol. 57, pp , 2016 [16]. de Almeda, Rogero G., JA Peças Lopes and J. A. L. Barreros, "Improvng power system dynamc behavor through doubly fed nducton machnes controlled by statc converter usng fuzzy control", IEEE Transactons on Power Systems, Vol. 19, No. 4, pp , 2004 [17]. Shuhu L, Tmothy A. Haskew and Lng Xu, "Control of HVDC lght system usng conventonal and drect current vector control approaches", IEEE Transactons on Power Electroncs, Vol. 25, No. 12, pp , 2010 [18]. Altyeb Altaher, Ammar Almoman and Sureswaran Ramadass, "Applcaton of Adaptve Neuro-Fuzzy Inference System for Informaton Securty, An Internatonal Journal of Computer Scence, Vol. 8, No. 6, pp , 2012 [19]. V. Calderaro, V. Gald, A. Pccolo and P. Sano, "A fuzzy controller for maxmum energy extracton from varable speed wnd power generaton systems", An Internatonal Journal of Electrc Power Systems Research, Vol. 78, No. 6, pp , 2008 [20]. Seung-Ho Song, Shn-l Kang, and Nyeon-Kun Hahm, "Implementaton and control of grd connected AC-DC-AC power converter for varable speed wnd energy converson system", In proceedngs of IEEE conference on Appled Power Electroncs, Vol. 1, pp , 2003 [21]. Joan Sau-Bassols, Eduardo Preto-Araujo, Samuel Galceran-Arellano and Orol Goms- Bellmunt, "Operaton and control of a Current Source Converter seres tappng of an LCC-HVDC lnk for ntegraton of Offshore Wnd Power Plants", An Internatonal Journal of Electrc Power Systems Research, Vol. 141, pp , 2016 [22]. Fernando D. Banch and Jos Lus Domnguez-Garca, "Coordnated Frequency Control Usng MT-HVDC Grds wth Wnd Power Plants", IEEE Transactons on Sustanable Energy, Vol. 7, No. 1, pp , 2016 [23]. Chayan Bhattacharjee and Bnoy Krshna Roy, "Advanced fuzzy power extracton control of wnd energy converson system for power qualty mprovement n a grd ted hybrd 34

22 Sunny Vg, et al. generaton system", IET Generaton, Transmsson and Dstrbuton, Vol. 10, No. 5, pp , 2016 [24]. Ast Mohanty, Sandpan Patra and Prakash K. Ray, "Robust fuzzy-sldng mode based UPFC controller for transent stablty analyss n autonomous wnd-desel-pv hybrd system", IET Generaton, Transmsson and Dstrbuton, Vol. 10, No. 5, pp , 2016 [25]. Rong Zeng, Le Xu, Langzhong Yao, Stephen J. Fnney and Y Wang, "Hybrd HVDC for Integratng Wnd Farms wth Specal Consderaton on Commutaton Falure", IEEE Transactons on Power Delvery, Vol. 31, No. 2, pp , 2016 [26]. M. Lnus and P. Damodharan, "Maxmum power pont trackng method usng a modfed perturb and observe algorthm for grd connected wnd energy converson systems," IET Renewable Power Generaton, Vol. 9, No. 6, pp , 2015 [27]. Al Ajam and Mehd Armaghan, "Fxed speed wnd farm operaton mprovement usng current-source converter based UPQC", An Internatonal Journal of Energy Converson and Management, Vol.58, pp.10 18, 2012 [28]. Ehab S. Al, "Speed control of nducton motor suppled by wnd turbne va Imperalst Compettve Algorthm", An Internatonal Journal of Energy, Vol.89, pp , 2015 [29]. Abdolvahhab Fetanat and Ehsan Khorasannejad, "Sze optmzaton for hybrd photovoltac wnd energy system usng ant colony optmzaton for contnuous domans based nteger programmng", An Internatonal Journal of Appled Soft Computng, Vol.31, pp , 2015 [30]. A.A. Mousa, Wael F. Abd El-Wahed and R.M. Rzk-Allah, "A hybrd ant colony optmzaton approach based local search scheme for multobjectve desgn optmzatons", An Internatonal Journal of Electrc Power Systems Research, Vol.81, pp , 2011 [31]. J. Wang, J. Cao, B. L, S. Lee and R. S. Sherratt, "Bo-nspred ant colony optmzaton based clusterng algorthm wth moble snks for applcatons n consumer home automaton networks," IEEE Transactons on Consumer Electroncs, Vol.61, No.4, pp , 2015 [32]. M. Ramesh and A. Jaya Laxm, "Enhancement of Power Transmsson Capablty of Hvdc System Usng Facts Controllers", An Internatonal Journal of Advances n Engneerng and Technology, Vol.1, No.5, pp , 2011 [33]. Moawwad, M. S. El Mours and W. Xao, "Advanced Fault Rde-Through Management Scheme for VSC-HVDC Connectng Offshore Wnd Farms," IEEE Transactons on Power Systems, Vol.31, No.6, pp , 2016 [34]. Vjay Kumar and N.V. Srkanth, "A hybrd approach for optmal locaton and capacty of UPFC to mprove the dynamc stablty of the power system", An Internatonal Journal of Appled Soft Computng, 2016 [35]. Reza Taghav, Al Reza Sef and Hadar Samet, "Stochastc reactve power dspatch n hybrd power system wth ntermttent wnd power generaton", An Internatonal Journal of Energy, Vol.89, pp , 2015 [36]. Amn Khodabakhshan, Mohammad Reza Esmal and Mosayeb Bornapour, "Optmal coordnated desgn of UPFC and PSS for mprovng power system performance by usng mult-objectve water cycle algorthm", An Internatonal Journal of Electrcal Power and Energy Systems, Vol.83, pp , 2016 [37]. M. Ajay Kumar and N. V. Srkanth, "An adaptve neuro fuzzy nference system controlled space cector pulse wdth modulaton based HVDC lght transmsson system under AC fault condtons", Central European Journal of Engneerng, Vol.4, No.1, pp.27-38, 2014 [38]. Sedgheh Danesh, Rahman Farnoosh and Tahereh Razzaghna, "Fuzzy nonparametrc regresson based on adaptve neuro fuzzy nference system", An Internatonal Journal of Neuro computng, Vol.173, No.3, pp ,

23 Optmal Power Dspatch of WECS and UPFC wth ACO and ANFIS Algorthms [39]. Fard Hashem, Noradn Ghadm and Behrooz Sobhan, "Islandng detecton for nverterbased DG coupled wth usng an adaptve neuro-fuzzy nference system", An Internatonal Journal of Electrcal Power and Energy Systems, Vol.45, pp , 2013 [40]. Poura Sarhad, Behrooz Rezae and Zahra Rahman, "Adaptve predctve control based on adaptve neuro-fuzzy nference system for a class of nonlnear ndustral processes", An Internatonal Journal of the Tawan Insttute of Chemcal Engneers, Vol.61, pp , 2016 Sunny Vg obtaned hs Bachelor s degree Electrcal Engneerng from Unversty of PTU. Then he obtaned hs Master s degree n Electrcal Engneerng from Punjab Engneerng College (Deemed to be Unversty), Chandgarh. Currently, he s pursung hs PhD from Punjab Engneerng College (Deemed to be Unversty), Chandgarh. Surjan Balwnder Sngh s Professor n the Electrcal Engneerng Department, Punjab Engneerng College (Deemed to be Unversty), Chandgarh. The author receved B.E. (Electrcal) n 1989, M. Tech. (Power Apparatus & Systems) n Feb 1991, and Ph.D. degrees n 2008, from Shvaj Unversty Kolhapur, I.I.T. Bombay, and Panjab Unversty Chandgarh respectvely. He has twenty years of professonal teachng experence n the same nsttute. He has taught undergraduate and post graduate students. He has guded number of post graduate and PhD students to complete ther thess manly n the feld of power system stablty studes also n feld of photometrc analyss of lumnares Hs areas of nterest nclude power system stablty studes, llumnaton engneerng, machne applcatons, modelng and analyss. The author s member of professonal socetes lke IEEE, Indan Socety of Lghtng Engneerng (M), Fellow Insttuton of Engneers (I), Chartered Engneer IE (I). 36