1 Euopean Aociation fo the Development of Renewable Enegie, Envionment and Powe Quality (EA4EPQ) Intenational Confeence on Renewable Enegie and Powe Quality (ICREPQ 2) Santiago de Compotela (Spain), 28th to 3th Mach, 22 Expeimental DTC of an Induction Moto Applied to Optimize a Tacking Sytem B. Mokhtai, A. Ameu, M. F. Benkhoi 2, L. Mokani and B. Azoui 3 LEDMaSeD Laboatoy, Electical Engineeing Depatment, Laghouat Univeity, BP 37G, Ghadaia Steet, Laghouat (3), Algeia, 2 IREENA, Polytech'Nante, Site de Saint-Nazaie, 37 bd de l'univeité - BP 46, 4462 Saint-Nazaie cedex 3 LEB Laboatoy, Electical Engineeing Depatment, Batna Univeity, Chahid M.E.H. Boukhlouf Steet, Batna (5), Algeia, Abtact. In ecent yea, the aea of indutial application of high pefomance AC dive, epecially the induction moto (IM), baed on Diect Toque Contolle (DTC) technique ha gadually inceaed due to it advantage ove the othe technique of contol. Among thee application ae cited, populion and tacking ytem. Thi wok peent the expeimental pat of a eeach poject conducted by the autho whee they have implemented an algoithm fo a DTC in ode to achieve a tacking ytem deigned fo a eie of ola panel to maximize it oveall efficiency. In thi pape, thee diffeent witching table ae peented fo chooing the bet between them fo a le fluctuation in toque. Thi tudy i jutified by educed cot of implementation compaed to othe method that equie the ue of expenive hadwae and complicated contol technique. The tet wee caied out unde the ame condition with the ame equipment of the tetbed, to get a good compaative tudy. Thee vaiant wee teted: imple witching table, table of ecto hifted by 3 and a table with 2 ecto. The eult how an impovement, but a dilemma between the toque ipple and thoe of the tato flux. Key wod Induction Moto, Diect Toque Contol, Switching Table, Toque and Flux Ripple, Tacking ytem, ola panel efficiency.. Intoduction High dynamic pefomance of evo moto dive i indipenable in many application of today automatically contolled machine. AC evo moto contol ha attacted much attention ecently in the powe electonic field . The IM ha eveal advantage ove othe type of electic moto. The ecent development of the technology of powe electonic ha put in competition a few othe type of moto, which had difficulty of deign in the pat, uch a the pemanent magnet ynchonou moto, fo example. But pobably, it will tay, fo next yea, the mot dominant moto in indutial application, becaue of it techno-economic chaacteitic [-6]. DTC wa intoduced in 985 by Takahahi and Depenbock epecially fo the aynchonou and ynchonou machine , . The main advantage of DTC ae the implicity of the contol cheme and it uneponivene to paamete vaiation (except tato eito), . In the conventional DTC, the employment of the hyteei contolle, in ode to egulate the tato magnetic flux and toque, have by natue high toque ipple and vaiable witching fequency depending on peed, load toque and hyteei band. Thi lead to a difficulty to contol toque and flux at a vey low peed [-8]. Lacu, C.et al.in , have peented a modified DTC uing (SVM) fo an IM with fixed witching fequency and low ipple fo toque and flux. Thi ytem equie two popotional-integal (PI) contolle popely tuned at the ame time fo the bet pefomance. Matin, C. et al in  and Ciincione, M., et al. in  have popoed olution to educe the toque ipple, thi tategy elie on inceaing the numbe of vecto applied voltage, which can impove the ipple band of toque. The IM which i poweed by a thee-level invete ha the ame dynamic pefomance a thoe obtained with a two-level invete with low toque and flux ipple, but thi inceae the cot of implantation, due to the cot of the thee-level invete and alo the witching table become lage. The aim of thi tudy i to have a imple and le expenive ytem, to apply it to a tacking ytem which wa popoed by , and can be ued fo themal ytem o photovoltaic; thi equie that it focue on impovement, oftwae ide only. Fo thi, we mut keep the ame equipment of tetbed and the ame contol condition (gain of PI egulato, efeence value, hyteei compaato), and theefoe, only the witching table mut be changed. The pefomance of ola panel ha alway been
2 conideed to be inufficient and we have to make them moe efficient to upply a moe ignificant enegy fo low and medium powe ytem. Among eveal paamete that influence pefomance we may cite the iadiation of unlight pojected onto the ola panel and captued by photovoltaic cell. Seveal tudie have been made to impove the oveall efficiency and thu benefit the mot fom the obtained tanfomed enegy [2-5]. The unlight captue block popoed by the autho in , i updated and coected o that it i eady to achieve oon. Figue how a diagam of the new tuctue. It diffe fom the peviou model by the poition of the photodiode which i eponible fo the detection of the unie, becaue it i moe convenient. So we mut place it in the back of the ola panel o that it i diectly in font of unie. The eat way of un ay The ide of ola panel Shadow Tube ized accoding to the pefomance degadation Panel uface Sunie photodiode Rotation axi Two photodiode Fig.. A new tuctue of the unlight captue block The expeimental tudy examine a DTC applied to an IM fo thee diffeent witching table: imple witching table, table with ecto hifted by 3 and table with 2 ecto in ode to educe the ipple of the electomagnetic toque and flux. 2. Optimal poition capto Thi block i deigned by thee photodiode, two ae located on the font of the ola panel, thei ole i to enue the optimal location of the un duing the day, while the thid photodiode located in the back of the ola panel to detect unie when the panel i facing wet. One of the two photodiode included in a tube of pedetemined length (depending on the deied accuacy of degadation of the unlight) to detect the hadow when the un move along it tajectoy. The ytem will be locked when we each the two logic tate: ' 'and '', whee ' 'epeent an excitation of the photodiode and '' epeent the abence of excitation. The following table lit all poible tate that can occu duing the whole day of opeation. Table. Summay of the ytem popoed duing the day Event State (logical) Action (contol) Sunie Rotate to the Eat Ray Panel Lock (no otation) The Sun Move Rotate to the Wet Ray Panel Lock (no otation) Cloud / Sunet Lock (no otation) -Cloud miing Rotate to the Wet -Sunie Rotate to the Eat If we ymbolize (left to ight) the tate of the fit photodiode with 'A', and that of the econd with 'B' and then we ue 'C' fo the thid, we can ee fom the table, that the movement will begin in two way:' ' and ''. 3. IM model The tato and oto flux equation of IM can be witten in the efeence fame of Pak in the following fom: Such a: A = B = X σ σl = σl T ω X = AX + BU t t [ i i φ φ ],U = [ v v ] α + β T α β - ω + σ T T α β σl T ω σl () ω σl σlt 2 m L L, σ =,T =,T = (2) L L In addition, the electomagnetic toque can be expeed by: T e 3 = p( φ α iβ -φβiα ) (3) 2 The mechanical equation of the IM can be expeed a flow:. J Ω = T T f Ω (4) 4. Conventional DTC e The DTC, a hown in figue 2, conit of diectly contolling the invete witche tun OFF o ON, on the calculated value of the tato flux and toque fom elation (6) and (7). The efeence fame elated to the tato, make poible to etimate flux and toque on the one hand and the poition of flux tato on the othe hand. The aim of the witche contol i to give the vecto epeenting the tato flux the diection detemined by the efeence value.
3 The ynthetic equence can be illutated though the following example: Auming that the flux vecto i located in ecto , then if the eo between the efeence flux and the tato flux i poitive, we mut inceae the flux, thi i only poible by applying a voltage vecto in the ame diection, accoding to (4) o V (), V2 () o V6 (). Howeve, applying voltage of oppoite diection V3 (), V4 () o V5 () deceae the vaiation of the flux. On the othe hand, if the eo between the efeence toque and the electomagnetic toque i poitive we mut inceae the electomagnetic toque by applying the voltage vecto in the half plane of poitive angle, accoding to (5), i.e. V2 (), V3 () o V4 (). Tying vecto V (), V5 () o V6 (), deceae the toque . t φ α = ( v α I α ) dt t φ β = ( ) v β I β dt (5) φ ( k + ) φ ( k) + VTE φ VTE (6) ' ' T = k ( φ + φ ) = k φ φ in( δ) (7) whee: T E i the ample time e Fig.2. Block Diagam of a Conventional DTC A two level claical invete can achieve even epaate poition in the phae coeponding to the eight equence of the voltage invete , . Table 2 how the equence fo each poition. Futhemoe, Table 4 and 5 have the equence coeponding to the poition of the tato flux vecto in diffeent ecto fo the othe tategie of ST (ee figue 3 and 4). The flux and toque ae contolled by two compaato with hyteei. The dynamic toque, ae geneally fate than the flux. So, a compaato hyteei of eveal level i, jutified to adjut the toque and minimize the witching fequency aveage . Table 2. Table genealized of voltage vecto geneated by a conventional DTC. Inceae Deceae φ V k-,v k and V k+ V k-2,v k+2 and V k+3 5. DTC with ecto hifted by 3 The ame pinciple of baic DTC contol of the IM, which i upplied by a two-level invete, i applied fo the othe ST. All ecto ae hifted by an angle of 3 leading to an angle of the fit ecto between and 6 a hown in figue 3. β V 4 =( ) V 3 =( ) S 3 S 4 S 2 S 5 V 2 =( ) S S 6 V 5 =( ) V 6 =( ) Secto V =( ) Fig.3. Diffeent vecto of tato voltage, in cae of table with ecto hifted by 3. The diffeence between a DTC with a imple witching table and that with ecto hifted i ummaized in table 3. Table 3. Compaion between the imple table and the table with ecto hifted by 3. Simple Table Secto hifted by 3 V toque undetemined DT, IF V IT, IF IT, IF V IT, DF flux undetemined V toque undetemined IT, DF V DT, DF DT, DF V DC, IF flux undetemined whee: I(D)T(F) : Inceae (Deceae) of Toque (Flux) Table 4 peent the witching table when the ecto ae hifted by 3. We can ee the diffeence fom the conventional DTC in the thid and the fouth line fo conventional witching table. Table 4. Switching table of DTC with hifting 3 φ T e S S 2 S 3 S 4 S 5 S DTC with 2 ecto The 2 ecto method ue the ame block diagam a hown in figue 2; but the witching table now conit of 2 non null voltage vecto, to be elected. The flux angle now lie on one of the 2 ecto a hown in figue 4. α
4 V 3 =( ) β S 4 S 3 V 2 =( ) Secto The cae of ecto hifted by 3 peent the mot emakable impovement. S 5 S 2 V 4 =( ) S 6 S 7 S S 2 V =( ) α S 8 S S 9 S V 5 =( ) V 6 =( ) V 3 (DF,IT) V 2 (IF,IT) Fig.6. Toque epone, fo all tategie of DTC. V4(DF,SIT) V 5 (DF,DT) V (IF,SDT) V 6 (IF,DT) Fig.4. Diffeent vecto of tato voltage in cae of table with 2 ecto. with: SI(D)T : Small Inceae (Deceae) of Toque. The witching logic i imila to that of a conventional DTC . Table 5 how the tategy fo the fit ecto and the 2 ecto whee we can deduce the et of all ecto. Figue 7 how the eaon why the conventional and the DTC with 2 ecto ae not o diffeent, which i confimed by . DTC with ecto hifted by 3 i well chaacteized by a diffeent PWM, which give bette impovement in tem of ipple. Table 5. Switching table of DTC with 2 ecto S S 2 Flux Toque Inceae V,V 2 and V 6 V,V 2 and V 6 Deceae V 3,V 4 and V 5 V 3,V 4 and V 5 Inceae V 2,V 3 and V 4 V,V 2 and V 3 Deceae V 5,V 6 and V V 4,V 5 and V 6 7. DTC expeimental eult A. Pat The paamete of the IM ued in thi tudy and the deciption of the tetbed ae given at the end of thi pape. Fo a compaion tudy, the eult ae gouped into et of thee. In Figue 5, the oto peed eache it efeence with a gap of about 4ad / fo the imple table and the table of 2 ecto, the efeence of thi peed i ad /. thi diffeence i due to the load which i 5 N.m. In the ame tet condition, the oto peed dop by about 2 ad / in the cae of ecto hifted by 3 compaed to the othe table, but it may be noted that fluctuation in peed, ae le peent in the latte tategy compaed to the othe two tategie. Fig.7. Voltage of a tato phae, fo all tategie. In figue 8, the tato flux ha fewe ipple in the cae of conventional DTC than the othe tategie. Thi cae peent a dilemma fo choice between ipple of toque and ipple of flux. The flux otation fequency, in the cae of ecto hifted by 3 i bigge than that of othe table; theefoe the flux pefom well in thee two table than in the table with ecto hifted, thi can intepet the ipple of tato flux. Fig.8. Stato flux, fo all tategie of DTC. Fig.5. Roto Speed, fo all tategie of DTC. The toque ha been impoved in the lat two tategie (Figue 6), but even if it wa chaacteized by annoying fluctuation. Fig.9. Occupation zone, fo all tategie.
5 B. Pat 2 In the following wok, we will chooe a cenaio fo teting the contol ytem, thi cenaio i ummaized in table 6. Table 6. Scenaio imulation fo tacking of ola panel. Peiod () Event State (logic) Action (movement) - 2,2 The Sun move Rotate to the Wet 2,2-4,62 Optimal poition Lock (no movement) 4,62-7,2 Sunie Rotate to the Eat 7,2-9,62 Optimal poition Lock (no movement) 9,62- The Sun move Rotate to the Wet Whee thee ae cloud, i imila to the unet, howeve, the tate that follow take '' indicating that the panel hould tun to the Wet. 8. Concluion In thi wok, we have examined and validated the algoithm of a imple DTC, applied to a tacking ytem in the aim to optimize a global panel ola efficiency. Thee witching table have been peented and teted; the expeimental eult how that the table with ecto hifted by 3 can impove the toque ipple which i inteeting fo the application in iue (tacking ytem), but not thoe of the tato flux. Table with 2 ecto can alo impove the contol lightly, confiming the eult of , but it epoted le inteet ince it employ a witching table lage enough compaed to the othe two tategie. Thi expeimental tudy help u to adopt the bet imple witching table, it i ecommended to ue the hifted table tategy to implement the tacking ytem deigned fo ola ytem (themal o photovoltaic). Deciption of the tetbed Fig.. Roto Speed, fo DTC with ecto hifted by 3 PC: P3(X86 Family 6, Model 8, tepping 6), 866 MHz, 256 MO (Ram), VGA: Matox Millinium G45 Dual head (32 MO), OS: Window 2 po. DSPACE: ISA, DSP 3 PPC Contolled Boad. INVERTER: Two level (max KHz), IGBT (2V- 5A). SENSORS:, Speed: Univeal DIGISINE, DHO5 [BEI-IDEACOD], 2, Cuent: AM3N -A/V, (Input: A peak max, DC. khz output:-mv/a, V peak max. 3, Voltage: Diffeential DP with two output - Fig.. Toque, fo DTC with ecto hifted by 3. whee: Fig.3. Tetbed ued in the peent tudy. : the IM 2: DSPASE conol 3: the two-level invete 4: the load (PMSG) Fig.2. Stato flux, fo DTC with ecto hifted by 3. To maintain the toque equalling zeo, one ha to cancel the tato flux, othewie one i not able to have a zeo toque in the moment whee the flux and the cuent ae not indicating zeo, becaue the unloaded cuent i till non null accoding to (2). Table 7. IM paamete ued in thi tudy Pole pai 2 Rated powe kw (at 5 Hz) 4 Rated voltage (V) 22/38 R (Ω).3 R (Ω).9 L, (H).9 M (H).8 J(kg.m²).9 f (N.m./ad).3
6 Acknowledgment Thi expeimental tudy wa conducted within the laboatoy IREENA in St Nazaie, Univeity of Nante. The IREENA' team membe find hee ou appeciation and acknowledgment. Refeence . I. Takahahi, and Y. Ohmoi, High-pefomance Diect Toque Contol of an Induction Moto. IEEE Tanaction on Induty Application, Vol. 25, Mach/Apil 989, pp . B.K. Boe, High Pefomance Contol of Induction Moto Dive. IEEE, IES Newlette, Vol. 45, pp. 7-, Sept . X. Roboam, B. De Fonel and M. Pietzak-David, Loi de commande diecte de couple du moteu aynchone, chapite 6, Modéliation Contôle Vectoiel et DTC, commande de moteu aynchone (Law of diect toque contol of aynchonou moto, Chapte 6, Modeling, Vectoial Contol and DTC, contol of aynchonou moto.) Vol., Hemè Science Euope 2 Edition. . G.M.A. Sahan, A.A. Nafeh, M.H. Shalan, Diect Toque Contol of PM Synchonou Moto. In Jounal of Electic Engineeing, Vol., 2, Edition 4, p . B. Mokhtai, A. Ameu, L. Mokani, B. Azoui and M.F. Benkhoi, DTC Applied to Optimize Sola Panel Efficiency. In Poc, IECON O9, 2-5 Novembe 29, Poto, Potugal, pp . L. Zhong, M.F. Rahman, W.Y. Hu, K.W. Lim, and M.A. Rahman, A Diect Toque Contolle fo Pemanent Magnet Synchonou Moto Dive. IEEE Tanaction on Enegy Conveion, Vol. 4, Septembe 999, pp . A. Ameu, B. Mokhtai, L. Mokani, B. Azoui, N. Eounbouli and A. Hamzaoui An Impoved Sliding Mode Obeve fo Speed Senole Diect Toque Contol Of PMSM Dive With a Thee-Level NPC Invete Baed Speed And Stato Reitance Etimato. In Jounal of Electic Engineeing, Vol., 2, Edition 4, pp.-9. . C.G. Mei, S.K. Panda, J.X. Xu and K.W. Lim, Diect Toque Contol of Induction Moto-Vaiable Switching Seno. Conf. Rec. IEEE-PEDS, July 999, Hong Kong, pp . C. Lacu, I. Boldea and F. Blaabjeg, A Modified Diect Toque Contol fo Induction Moto Senole Dive. IEEE Tan. Ind. Applicat, vol. 36 no., Jan/ Feb 2, pp . C. Matin, X. Roboam, T.A. Meynad and A.S. Cayalho, Switching Fequency Impoition and Ripple Reduction in DTC Dive by Uing a Multilevel Convete. IEEE Tan. Powe Electon., vol. 7, iue 2, Mach 22, pp . M. Ciincione, M. Pucci and G. Vitale, A novel diect toque contol of an induction moto dive with a thee-level invete. IEEE Bologna Powe Tech Confeence, June 23-26, Italy, 23, Vol. 3, pp.7. . E. Hoain, R. Muhida, and A. Ali,, Efficiency impovement of ola cell uing compound paabolic concentato and un tacking ytem, IEEE Electical Powe & Enegy Confeence (EPEC 28), Vancouve, Canada, Octobe 6-7, 28. . C. Alexandu, M. Comit, The enegy balance of the photovoltaic tacking ytem uing vitual pototyping platfom. Euopean Electicity Maket, 28. EEM 28, 5th Intenational Confeence on, 28-3 May 28 . R. Lau, H. Kim; M. Pang, A. Neidhadt, A. Cineo, V. Kaul, Self-coecting Adaptive Tacking Sytem. Militay Communication Confeence, 28. MILCOM 28, IEEE, 6-9 Nov. 28, pp.-7. . C. Jung-Sik, K. Do-Yeon, P. Ki-Tae, C. Chung-Hoon, C. Dong-Hwa, Deign of Fuzzy Contolle baed on PC fo Sola Tacking Sytem. Smat Manufactuing Application, ICSMA 28. Intenational Confeence on, 9- Apil 28, pp . B.S. Kuma, R.A. Gupta, and R. Kuma, 2-Secto Methodology of Toque Ripple Reduction in a Diect Toque Contolled Induction Moto Dive. SICE-ICASE Intenational Joint Confeence 26, Oct. 8-2, 26, in Bexco, Buan, Koea, pp