Optimized Fuzzy Controller Design to Stabilize Voltage and Frequency Amplitude in a Wind Turbine Based on Induction Generator Using Firefly Algorithm

Size: px

Start display at page:

Download "Optimized Fuzzy Controller Design to Stabilize Voltage and Frequency Amplitude in a Wind Turbine Based on Induction Generator Using Firefly Algorithm"

Chrystal Rodgers
5 years ago
Views:

Intenational Reseach Jounal of Management Sciences. Vol., 3 (3), 12-117, 215 Available online at http://www.ijmsjounal.

1 Intenational Reseach Jounal of Management Sciences. Vol., 3 (3), , 215 Available online at ISSN x 215 Optimized Fuzzy Contolle Design to Stabilize Voltage and Fequency Amplitude in a Wind Tubine Based on Induction Geneato Using Fiefly Algoithm Pouiya Khani Maghanaki, Abdolhossein tahini Depatment of Electical Engineeing, Islamic Azad Univesity, Damghan Banch, Damghan, Ian * Coesponding Autho: Pouiya Khani Maghanaki ABSTRACT: The use of wind tubines based on induction geneato is vey popula to geneate electical powe and it has been noted by eseaches because of its many advantages compaed to conventional methods of electical enegy geneation. Factos of uncetainty in the natue of the wind cause vaiable voltage in amplitude and fequency on the induction geneato. It is not appopiate to apply such voltage to the load. So a contolle must be designed to be kept constant voltage and fequency. In this pape, a fuzzy contolle is used as state feedback to stabilize the voltage, fequency and voltage amplitude. The vaiable AC voltage geneated by geneato is conveted by ectifie to a vaiable DC voltage. The vaiable DC voltage causes a change in the output voltage of the invete. The PWM switching popety is used to stabilize fequency and state feedback is used to stabilize the output voltage amplitude. The obtained eo signal with its deivative is applied to the fuzzy contolle as input to geneate the consideed contol signal by contolle to geneate appopiate fiing pulses to apply to PWM invete. Theefoe, fequency and amplitude of the output voltage is kept constant with switching contol and so maximum powe of wind is esulted. In this pape, the fiefly algoithm is used to optimize the paametes of fuzzy contolle and to impove its pefomance to stabilize the voltage and fequency. Simulation esults show that by design the appopiate contolle fo the consideed system output voltage can be stabilized in constant amplitude and fequency. The esults showed that the optimization of fuzzy contolle inceases its ability and pefomance in voltage amplitude and fequency stabilization. Keywods: Wind tubine based on induction geneato, Voltage and fequency amplitude stabilization, Fuzzy contolle, Fiefly algoithm. INTRODUCTION Pollution is one of the most impotant issues in the use of enewable esouces which gows inceasingly its impotance. So that it has foced intenational oganizations like the IEA (Intenational Enegy Agency) to take seious decisions to potect the health of the planet and educing atmospheic pollution. This can be a geat help to the seious foecasting to use enewable enegy souces. Wind enegy is cuently the fastest enegy fom point of view of spead in the wold. Today, wind enegy is gowing at a 3 pecent gowth in wold (Mulle et al., 22). In tems of envionmental, wind powe not only educes the poduction of cabon dioxide which is the main cause of geenhouse gas emissions, but it hasn't othe pollutions esulting fom the use of fossil fuels (Lisee et al., 211). In tems of wind powe geneation costs, these costs ae deceasing day by day so that the cost of each KWh powe geneated by wind tubine had deceased 2 pecent duing 5 yeas. Also in tems of size wind tubines come to maket with highe powes while these tubines have an output of 5 MW (Petesson et al., 25). The use of wind powe plants with vaiable speed has advantages compaed to the fixed speed wind powe plants. Although wind powe plants with a constant velocity, can be connected diectly to the netwok, howeve, a wide ange of enegies is coveed by the vaiable speed wind powe plants and has less mechanical stess and noise. Today, with advances in powe electonics, all speeds contol is possible and effective (Pena et al., 1996). In tubines with vaiable speed, in fact, the otay pat of tubine absobs the mechanical powe fluctuations with changes in its speed and output powe cuve is flatte. This can help to impove the quality of powe. Howeve, wind is a vaiable quantity and evey moment is changing. So the voltage amplitude and output fequency of the wind tubine is changing that vaiable voltage is not suitable fo consume. So a contolle must 12

2 be designed to be capable to delive maximum voltage and powe with constant amplitude to consume fom tubine. In this aticle, the fuzzy contolle is designed to maintain the amplitude and fequency of available voltage caused by the uncetainty of natue wind. In this aticle, the fuzzy contolle optimized by the fiefly algoithm is used to maintain the amplitude and fequency of available voltage caused by the uncetainty of natue wind. Studied system The studied system consists of a wind tubine is connected to the load as shown in Figue 1.The system includes a wind tubine, induction geneato, diode bidge ectifie, filte DC, invete and suitable AC filte. Figue 1. Oveview of a wind tubine connected to the load. Figue 2 shows a geneal view of a wind tubine block, the adaptive contolle and the induction geneato. Figue 2. A geneal view of a wind tubine block, the adaptive contolle and the induction geneato. Wind tubine Toque poduced by the wind tubine with no losses is as follows (Raina & Malik, 1983): T C ( ) R V )1 ( a t 2 13

3 R V C C p t )2 ( )3 ( Whee R is the adius of the wind tubine, is the tubine otation speed and is the atio of tip speed to the wind linea speed. Adaptive contolle as feed fowad Due to changes in the paametes in studied system, using the fixed gain classic contolles is not easonably fixed. Because this contolle is not capable to maintain system stability in conditions of paametes vaiation. To solve this poblem, the adaptive contolle can be used. Theefoe, this contolle is used in wind fam. The oveall block diagam of the adaptive contolle is shown in Figue 3. Figue 3. The oveall block diagam of the adaptive contolle. Calculation of * * is obtained fom following equation (Raina & Malik, 1983): * n opt m V R )4( In the above equation n=98 and R= 33 is the tubine adius and model of * is consideed as a gain in wind speed. Contolle The equation of the contolle is consideed as follows: v * m ke The value of Ke is 1.8 in the above equation. Adaptive computing of Tˆ t Accoding to the adaption law, we have (Raina & Malik, 1983): T e( t) H( s) k ( t) v( t) (6) In the above equation e (t) is a scala quantity, H(s) is the eal tansfe function, k is the constant with given sign, is the 1*M vecto function of time and V(t) is the measuable vecto of 1*M. If is changed as follows (Raina & Malik, 1983): ( t) sgn( k) ev( t) )7( With the constant and positive, the e(t) and ae bounded. If V is bounded, so: e( t) as t )8) Now with the last equation fo the dynamic eo, we have following equations: 1 ~ (9( e ke Tt J R Fom the above equations, the esult is the following adaptation law: (5) 14

4 Tˆ t e J R is a constant value that is consideed 2. So the J R 2 1 J n t J In the above equation: J J T M n 98 6 M 63.87Kgm Kgm Feedback lineaization (T em * calculation) 2 2 (11) J R is defined as follows: (12( A toque which is estimated by an adaptive contolle fo geneatoand is applied as input to it, is obtained by following equation: T (13( ˆt Tem J R v J R Simulated adaptive contolle model Given the above steps,adaptive contolle simulation model is pesented in Figue 4. Induction geneato Figue 4. Simulated adaptive contolle. The Pak tansfom is used in induction geneato model. In this tansfom of thee-dimensional space (a-b-c) is tansfomed to the (d-q-).geneato input equations has become to its new fom by Pak tansfom. Blocks and new equations ae as following. Also the outputs ae (a-b-c) by the invese Pak tansfom (Hu & He, 26; Zhou et al., 24). d v (14) ds sids ds dt d v i (15) v v s v d q s i s os d os dt d ds dt i ( ) i ( ) q v i T em d dt 3P ( dsi i 4 q d ds d d dt d q dt ) (16) (17) (18) (19) (2( 15

5 L Rs La ds Lm R d La q L R L a s Lm R L a LmR L a a LsR L s L ds mr v s L v a d v q v LsR L a ds d q (21) Equations of induction geneato ae as follows (Hu & He, 26; Zhou et al., 24): ds os b b b s v ds ( mq ) d b xls s v ds ( md ds) dt b xls (v i ) d os s os t q b v q ( ) d ( mq q) dt b xl mq i xls ds md ids xls q mq iq xl d md id xl q mq xm ( ) xls x l ds d md xm ( ) xls xl 3 P Tem ( dsi ids) 2 2b b ( e L ) 2 t (22) (23) (24( )25( (26) (27) (28( )29( )3) (31) (32( (33( Block cicuits of d and q axis The block of the q axis and inside of the q axis block is shown by Fig. 5. Figue 5. Inside the q axis. The block of the d axis and inside of the d axis block is shown by Fig 6. 16

6 The block cicuit of the oto and abc to dqo tansfom Figue 6. Inside the d axis. The inside of oto blocks and abc to dq tansfom is shown by Figues 7 and 8. Figue 7. Inside the oto block. 17

7 Applying filtes Figue 8. Inside the abc to dqo tansfom block. AC/DC and DC/AC convetes ceate hamonics fom AC and DC side that may have no suitable size fo the souce o load. The filtes ae used to educe the hamonics. Filte that is used in the AC side ae to educe the cuent hamonics and on eithe side of the DC is to educe the voltage hamonics amplitude. The fuzzy contolle design The fuzzy contolle uses the fuzzy logic ules to obtain contol applications. Fuzzy ules have been established based on contol ules. Fuzzy logic systems ae not designed based on mathematical models. Fuzzy contolles implement the human logic using fuzzy logic that is planned by membeship functions, fuzzy ules and membeship ules. In this study, the output voltage is sampled and is compaed with the efeence voltage level and the geneated eo signal and its deivative is applied to the contolle as the contolle inputs. A fuzzy system is descibed by a set of IF-THEN ules and uses a numbe to define the degee of membeship in its membeship functions. To solve the poblem in fuzzy contolle, contolle inputs, eo signal and its deivation, and its output ae consideed as the contol signal. The wind tubine system with fuzzy contolle stuctue is shown in Fig

8 Figue 9. The Poposed fuzzy type-1 contolle in wind tubine system. Design of fuzzy contolle has 4 steps that ae pesented as follows (Zadeh, 1988): Step 1: detemination of the system dynamic behavio and chaacteistics In this step, the contolle input and output vaiables and thei vaiations ange consideing load-fequency contol poblem should be detemined. Input signals ae consideed to ceate the base ules as the IF pat and the output signal of the fuzzy contolle ae consideed as the THEN pat. Step 2: Detemination of the fuzzy sets and membeship functions In this step, the degee of fuzzy membeship functions elated to each input and output signals ae detemined and the fuzzification pocessing is completed. The fuzzy contolle membeship functions ae shown in Fig. 1. NB NS ZZ PS PB 1 Degee of membeship A x 1-3 S M B VB VVB 1 Degee of membeship K Figue 1. The membeship functions of poposed fuzzy contolle (A: input and K: output). 19

9 Step 3: Desciption of infeence engine In this step, fuzzy ules ae fomed by the contol ules that system pefomance is based on that ules. Step 4: fomation of the defuzzification pocess The fuzzy ules ae combined and defuzzification pocess is pefomed on the output and output of the fuzzy infeence engine. In this study the membeship functions inputs and output ae consideed same. Fo inputs and output, the 5- segments tiangula membeship functions ae applied. Each input has five membeship functions, so the numbe of base fuzzy ules is 25. The fuzzy ules of fuzzy contolle ae pesented in Table 1. Table 1. Fuzzy ules of poposed fuzzy type-1 contolle. ec e NB NS ZZ PS PB NB S S M M B NS S M M B VB ZZ M M B VB VB PS M M VB VB VVB PB B VB VB VVB VVB Optimized Contolle Based on Fiefly algoithm The fiefly algoithm is applied fo designing the self-tuning adaptive fuzzy contolle. Also the fiefly algoithm is used to achieve a bette pefomance of fuzzy contolle to optimize the membeship functions paametes with the aim of minimizing system fequency and voltage amplitude deviations. In the system unde study, ACE (eo signal) and ACE (deivation of eo) ae consideed as the input signals and the output signals ae used to change the set points in the poblem. Fiefly Algoithm Fiefly optimization method is a new evolutionay optimization method which was intoduced in 27 by M. Yang (Gandomi et al., 211).The main idea of this method is fom optical communication between fieflies. This algoithm can be seen as manifestations Swam intelligence, whee fom the coopeation (and possibly competing) of simple and less intelligent membes highe ode of intelligence is ceated that is definitely not obtainable by any of the membes. Fiefly Algoithm that is inspied by the behavio of the natual species to optimize the ugged non-linea functions is one of the most successful algoithms and is inexpensive at the same time. Thee ae two impotant points on the fiefly algoithm, Changes in light intensity and Fomulation of attactiveness. Fo simplicity we assume that the attactiveness of a fiefly, it is dependent on light intensity which, as mentioned the objective function is the same as the light of a fiefly. Theefoe we can wite: Ii F( xi ) )34) Othes must detemine the attactiveness of a fiefly. Because of its bightness deceases with distance fom the fiefly, it also educed its attactiveness which the following equation can be used to calculate the bightness fom pespective of adjacent fiefly. Iij ij I e Whee ij is distance between two vaiables which is detemined accoding to equation 12-4, γ is the light absoption coefficient and I is the light intensity at zeo distance. Usually the choice is between.1 to 1, the value of 1 is consideed in this thesis [13]. x x ( x x ) ij i j i, k j, k k1 d 2 Whee d is the numbe of vaiables of the optimization poblem. The attactiveness of each fiefly to adjacent fiefly, is based on the light intensity of that fiefly and is calculated accoding to equation 13-4as it is clea that its value deceases with inceasing distance. B ij ij Be (37) Whee B is attactiveness in zeo distance which geneally its value is consideed 1 in the papes.if a wam has a lowe bightness elative to its adjacent wam, it moves towad it accoding to following equation: new 1 x ij i xi Be ( and ) )38( 2 Whee α is a coefficient of Fiefly algoithm which in this thesis we assume that its value is 1/and the and is a andom numbe between and 1. As is clea, moving a Fiefly has thee pats. The fist mode is the cuent status of the fiefly. The second pat is the move towads a bighte o moe attactive wam and the thid pat epesents the andom motion of Fiefly. The Fiefly algoithm can be shown as step by step though the following: (35) (36) 11

10 Step 1: Definition of basic paametes such as numbe of fieflies (n), B, αandγ and the total numbe of iteations of the algoithm. Step 2: Geneate andom vaiables within defined ange of the size n, (x=xmin+(xmax-xmin*and(1,nv))) Step 3: Calculate the objective function fo each set of vaiables (I) Step 4: fo i=1:n Step 5: fo j=1:n Step 6: If the j-th Fiefly has an intensity geate than Fiefly i,i th fiefly will move accoding to equation Step 7: The set of i th vaiable is to be updated. Step 8: End j Step 9: End i Step 1: finish, be eligible to convege, go to step 3. Optimized Contolle Application Fo bette pefomance of poposed fuzzy contolle, the pocesso blocks of contolle (P1, P2 and P3) ae optimized in specified ange with the aim of minimizing system fequency and voltage amplitude deviations. The pocesso blocks used to nomalize the contolle fo bette pefomance in damping of system fequency and voltage amplitude deviations. The fuzzy contolle consideing pocesso blocks is shown in Fig. 5. Figue 11. The fuzzy contolle consideing pocesso blocks. The optimal and accuate pefomance of poposed fuzzy contolle depends on the selection of best membeship functions. The fiefly algoithm is applied to detemine the paametes of k1, k2 and k3 consideing the objective function of optimization poblem in Fig. 6. The objective function is chosen as the Integal Squae Eo (ITSE) as follow: t sim 2 2 Cost Fuction ( eo eo ) dt (34( d q METHODOLOGY The pesent study is pactical in tems of natue and pupose, and in tems of methodology and collection of desciptive data, it is a coelation suvey. Reseach population included the customes of Expot Development Bank of Golestan povince fom whom 13 people wee selected using simple andom. Fo data collection, standad SERVQUAL questionnaies wee used fo sevice quality and custome satisfaction. Validity of both questionnaies was confimed by expets. To make sue of the validity of the questionnaies, Conbach s alpha was used and validity coefficient fo sevice quality and custome satisfaction questionnaies wee obtained.78,.82,.89,.87 and.82 fo sevice qualities of, espectively, tangibles, eliability, esponsiveness, assuance and empathy. Accoding to the eseach topic and hypothesis, infeential statistics of Peason s coelation using SPSS softwae was applied fo data analysis. RESULTS The studied system data elated to the induction geneato is pesented by Table 2 (Slootweg et al., 23). 111

11 Table 2. Simulation paametes (Slootweg et al., 23) Paamete Rated Powe Line to line voltage R S R X S X X m Poles Numbe Value 1.6 MVA 33 KV Oveview of system simulated in Simulink is shown in Figue 12. In this section, a thee phase fault is applied to the system accoding to Figue 12 and with design of fuzzy contolle is attempting to fix and maintain the voltage and fequency fluctuations. Figue 12. Oveview of the simulated system. 112

12 Input voltage cuves of the studied system is pesented in Figue 13. Figue 13. Input voltage cuve of system. Rectifie and invete output cuves ae shown in Figs. 14 and 15, espectively. Figue 14. The output voltage of the ectifie. 113

13 Figue 15. The output voltage of the invete. Total hamonic disode of ectifie output voltage is shown in Fig 16. Figue 16. Total hamonic disode of ectifie output voltage. Output voltage cuve of the studied system is pesented in Figue 17. Total hamonic disode of system output voltage is shown in Fig

14 Figue 17. Output voltage cuve of the studied system. Figue 18. Total hamonic disode of system output voltage with fuzzy contolle. Accoding to the Fig 18, it is clea that oscillations of output voltage as possible is fixed and total hamonic disode of system output voltage is equal to 4.91 pecent. At this stage, vaiables of fuzzy contolle is detemined using the section 4-2 with fiefly algoithm with the aim of minimizing the oscillations as optimal. fiefly algoithm paametes ae summaized in Table 3. The fiefly algoithm convegence cuve is shown in Figue 19. Table 3. The fiefly algoithm optimization Paametes. β Paamete Value Numbe of fileflies 15 γ 1 1 α.2 Max iteation 4 115

15 1 5 Cost Function Iteation Figue 19. Convegence cuve of fiefly algoithm. Accoding to Figue 19, clealy, the objective function value afte iteation 8 is constant. Afte optimization, the optimized paametes of contolle in Table 4 wee obtained. It should be noted that the minimum and maximum value of vaiables is consideed zeo and 1. Table 4. The optimal values of fuzzy contolle vaiables. P1 P P3.61 Afte detemining the optimal paametes of fuzzy contolle using the fiefly algoithm, the simulation was epeated with the optimized vaiables of fuzzy contolle and by applying thee phase fault. Fo compaison with pevious esults fom the optimization of fuzzy contolle total hamonic disode of system output voltage with fuzzy contolle is pesented by Fig 2.Total hamonic disode of system output voltage in this stage is 4.65.While its value befoe optimization of fuzzy contolle was obtained equal 4.Accoding to the esults value of fluctuations is deceased 3.5 pecent. Thus, by optimization of fuzzy contolle system voltage fluctuations can be educed. Figue 2. Total hamonic disode of system output voltage with optimized fuzzy contolle. CONCLUSION In this pape, a fuzzy contolle is designed fo stabilization of the amplitude and fequency of the vaiable voltage caused by the uncetainty of the wind natue. A fuzzy adaptive contolle was designed as diect supply to use 116

16 maximum wind powe at each moment. As is clea fom the esults, by applying the eo associated with changes in wind speed, Adaptive fuzzy contolle employs wind maximum powe and then voltage was ceated in induction geneato. Induced voltage has vaiable fequency and amplitude which wee kept constant in amplitude and fequency by fuzzy contolle and powe electonics cicuits. Accoding to the esults output voltage is within the standad in point of view of THD. In Continue, fuzzy contolle paametes wee optimized using the fiefly algoithm to impove the pefomance of the contolle in oscillation damping. Compae simulation esults befoe and afte optimization showed that by optimizing contolle its pefomance is impoved in stabilization of voltage and fequency fluctuations while total hamonic disode of system output voltage is educed with espect to the befoe of optimization. REFERENCES Gandomi AH, Yang X, Alavi AH, 211. Mixed vaiable stuctual optimization using Fiefly Algoithm. Computes & Stuctues. 89(23 24): Hu J, He Y, 26. Dynamic modeling and obust cuent contol of wind- tubine used DFIG duing AC voltage dip. J. Zhejiang Univ. Sci. A. 7(1): Lisee M, Cadenas R, Molinas M, Rodiguez J, 211. Oveview of multi-mw wind tubines and wind paks. IEEE Tans. Ind. Electon. 58(4): Mulle S, Deicke M, De Doncke RW, 22. Doubly fed induction geneato systems fo wind tubines. IEEE Ind. Appl. Mag. 8(3): Pena R, Clae JC, Ashe GM, Doubly fed induction geneato using back-to-back PWM convete and is application to vaiable-speed wind enegy geneation. Poc. IEE B Elect. Powe Appl. 143(3): Petesson A, Hanefos L, Thiinge T, 25. Evaluation of cuent contol methods fo wind tubines using doubly-fed induction machines. IEEE Tans. Powe Electon. 2(1): Raina G, Malik O, Wind Enegy Convesion Using a Self-Excited Induction Geneato. IEEE Tans.PAS.Vol.12. Slootweg JG, de Haan SWH, Polinda H, Kling WL, 23. Geneal Model fo Repesenting Vaiable Speed Wind Tubine in Powe System Dynamics Simulations. Powe Sys. 18(1): Zadeh LA, Fuzzy Logic. Compute. 1(4): Zhou F, Joos G, Abbey C, Jiao L, Ooi B, 24. Use of Lage Capacity SMES to Impove the Powe Quality and Stability of Wind Fams. Powe Engineeing Society Geneal Meeting. 2:

where and are polynomials with real coefficients and of degrees m and n, respectively. Assume that and have no zero on axis.

where and are polynomials with real coefficients and of degrees m and n, respectively. Assume that and have no zero on axis. function whee is an unknown constant epesents fo the un-modeled dynamics The pape investigates the position contol of electical moto dives that can be configued as stuctue of Fig 1 This poblem is fomulated