ISSN Vol.04,Issue.07, June-2016, Pages:

Transcription

1 ISSN l.04,issue.07, June-206, Paes: Fuzzy Induced fficiency Optiization of Paralleled D-D Boost onverter H. SHRAVANI, K. HTASWI 2 PG Scholar, Malla Reddy nineerin ollee (Autonoous), India, -ail: shravanichalla92@ail.co. 2 Assistant Professor, Malla Reddy nineerin ollee (Autonoous), India, -ail: chetaswi@rec.ac.in. Abstract: In this paper we introduce a new topoloy for D- D boostin operation with N-parallel converters with current sharin capability. The current sharin reduces the stress on the power electronic switches with increase in reliability of the converter. The feedback control loop is dynaic and robust with fuzzy interface structure at the duty ratio. The chane in the input resistance chanes the current in each branch and all the dynaic responses are observed in MATLAB Siulink software with coplete raphical results. Keywords: D-D Boost onverter, Fuzzy, onverters. I. INTRODUTION In any technical applications, it is required to convert a set voltae D source into a variable-voltae D output. A D-D switchin converter converts voltae directly fro D to D and is siply known as a D onverter. A D converter is equivalent to an A transforer with a continuously variable turns ratio. It can be used to step down or step up a D voltae source, as a transforer. D converters are widely used for traction otor control in electric autoobiles, trolley cars, arine hoists, forklifts trucks, and ine haulers. They provide hih efficiency, ood acceleration control and fast dynaic response. They can be used in reenerative brakin of D otors to return enery back into the supply. This attribute results in enery savins for transportation systes with frequent steps. D converters are used in D voltae reulators; and also are used, with an inductor in conjunction, to enerate a D current source, specifically for the current source inverter. [] fficiency, size, and cost are the priary advantaes of switchin power converters when copared to linear converters. The switchin power converter efficiencies can run between 70-80%, whereas linear converters are usually 30% efficient. The D- D Switchin Boost onverter is desined to provide an efficient ethod of takin a iven D voltae supply and boostin it to a desired value. A ood reliability of enery conversion is always needed. By dianosin the fault in its early staes, the reliability of the power conversion syste can be increased sinificantly [2]. To achieve this, it is very interestin to develop efficient online paraeter estiation. Many paraeter estiation techniques have already been proposed for dc dc boost converters. urrent sharin is an iportant functionality for parallel power converters to ensure reliable and efficient operation [8]. A ood review of current-sharin techniques has been proposed by hen et al. in [9] on different interconnection schees. A first ethod for controllin parallel boost converters is to define one converter as the aster which iposes output voltae while others are slave and only current reulated. Unfortunately, this control stratey is not suitable for a ood reliability of the lobal syste since it iht suffer fro fault tolerance aainst the failure of the aster converter []. Another ethod found in the literature is to use current sharin to anae the output voltae error as done in [] by usin syneretic control. Fi.. N parallel boost converters structure with one output capacitor. In a boost converter, the output voltae is reater than the input voltae hence the nae boost. A boost converter usin a power MOSFT is shown below Fi.2. Fi. 2. Sinle boost converter. 206 IJIT. All rihts reserved.

2 The function of boost converter can be divided into two odes, Mode and Mode 2. Mode beins when transistor M is switched on at tie t=0. The input current rises and flows throuh inductor L and transistor M. Mode 2 beins when transistor M is switched off at tie t=t. The input current now flows throuh L,, load, and diode D. The inductor current falls until the next cycle. The enery stored in inductor L flows throuh the load as shown in Fis.3 and 4. Fi. 3.hare ode. H. SHRAVANI, K. HTASWI represent the overall losses throuh the structure. For a ore detail description of this loss odelin technique, the reader is invited to read reference [7], where analytical study is presented, as well has load dependence behavior of such a odelin in the case of a sinle-boost converter.it has to be noticed that the estiation of losses throuh the converters can also been used for others purpose. As an exaple, it can be used for verifyin the aein of the individual converters by checkin on the variations of the estiated equivalent resistors. This aspect cannot be developed in this paper since it is not its oal, but reader is aain invited to read [7] for ore details on the estiation. III. ONTROL STRUTUR xponential stability can be proved easily with the classical Lyapunov approach [7]. The stability and dynaics of the proposed estiations are deterined by coefficients λ sk and λ p. In order to deonstrate their stability, the Lyapunov candidate Fi. 5. ontrol stratey for on parallel converter. Fi. 4. Dischare ode. II. STRUTUR MODLING AND STIMATION A. Model of the Structure The syste consists in N parallel boost converters with one output capacitor as shown in Fi. In this paper, it is proposed to odel losses throuh a boost converter by addin N serial and one parallel resistors in the conventional ideal odel. For the considered application, each individual boost converter odel has a serial resistance (rsk), while a unique parallel resistor R p includes all the rest of the losses for the whole structure. This difference with the ethod proposed in [7] coes fro the nature of the structure. ven if the adopted loss odelin is only represented throuh resistors, it is useful to underline that not only ohic losses are taken into account. In fact, every loss throuh the converter is taken into account with those of equivalent resistors, such as core hysteresis and eddy current losses, conduction ohic losses, and switchin losses of seiconductors [7]. Particularly, the parallel resistor R p does not only represent the capacitor o losses throuh its SR. Indeed, it is well known the even under zero power, boost converters still present losses, which will be taken into account throuh paraeter R p while serial resistor is not able to odel this behavior. Finally, paraeters r sk and R p () It has been decided to indirectly control the output voltae V o by reulatin the enery () stored in the capacitor o. As detailed in [7], it is possible to deonstrate that is a flat output for the considered syste as shown in Fi.5. (7) To ensure current reulation, it has been decided to desin a slidin-based controller. Slidin based control allows robust (2) (3) (4) (5) (6)

3 Fuzzy Induced fficiency Optiization of Paralleled D-D Boost onverter reulation. It has already been applied on the case of paralleled boosts where robustness aainst inductor variations is underlined and uaranteed. For each individual boost converter. For siulations and experiental validation, the switchin frequency is set to fs =20 khz. In a two-loop control schee, dynaics of the reulation ust be separated, and then ωn is chosen in order to satisfy (8) Then, a new ethod to split the power between the parallel boost converters is presented. The proposed repartition ais to axiize the lobal efficiency of the structure. Indeed, it is possible to deonstrate that there is a unique point which iniizes the total input power reference P refin. As an exaple, in this case the iniu value differs fro equal repartition possibility. Then, the interest of the proposed ethod appears easily copared to the ost used currentsharin technique. Therefore, it is needed to deterine the repartition paraeters αk, so that the input power reference P refin is iniu for a iven load reference P refout. Fi.7. Verify that 3 are selected as the Nuber of MFs. lick OK to add three Gaussian curves to the input variable service.. Renae the ebership functions for the input variable service, and specify their paraeters. lick on the curve naed f to select it, and specify the followin fields in the urrent Mebership Function (click on MF to select) area: In the Nae field, enter poor. In the Paras field, enter [.5 0].The two inputs of Paras represent the standard deviation and center for the Gaussian curve. Fi.6. The process of specifyin the ebership functions for the two input tippin exaple, tipper, is as follows: Double-click the input variable service to open the Mebership Function ditor. In the Mebership Function ditor, enter [0 0] in the Rane and the Display Rane fields. reate ebership functions for the input variable service. Select dit > Reove All MFs to reove the default ebership functions for the input variable service. Select dit > Add MFs to open the Mebership Functions dialo box. In the Mebership Functions dialo box, select auss f as the MF Type. Fi.8. lick on the curve naed f2 to select it, and specify the followin fields in the urrent Mebership Function (click on MF to select) area: In the Nae field, enter ood.

4 - I In the Paras field, enter [.5 5]. lick on the curve naed f3, and specify the followin fields in the urrent Mebership Function (click on MF to select) area: In the Nae field, enter excellent. In the Paras field, enter [.5 0]. The Mebership Function ditor window looks siilar to the followin fiure. IV. SIMULATION RSULTS Siulation on three-parallel boost converters has been realized to verify the effectiveness of the proposed ethod. First, the current repartition is set to αk = /3 At tie t =0.5 s, the proposed power repartition is enabled. Serial and parallel resistors of the siulated odel are chosen constant for an easier coputation. To validate our purpose, the siulated paralleled boost converters on which the proposed control is applied present different efficiencies for each individual converter. This is siulated by iposin rs =0.39 Ω,rs2 =0.39 Ω, and rs3 =.40 Ω traducin ore losses throuh the third individual converter. This difference between serial resistors is only one odelin of a poorer behavior of the third converter. On a real boost converter, any reasons can lead to such results, as any different losses are taken into account throuh this resistor. First, it is ensured that the estiated resistors convere to their siulated values. The calculated optial repartition coefficients as αopt =0.439 αopt2 =0.439 αopt3 =0.22. This results on different current references for each converter. Fi. 9 shows current on the previously described control schee. Fi. 0 shows the efficiency of the structure. In this fiure, the efficiency of the proposed current-sharin schee can be observed. Indeed, when total power is equally divide throuh eleents (t< 0.5 s), efficiency is about 2.4% less than the efficiency with the proposed ethod. Note that the difference with respect to experiental result coes fro the hypothesis that estiated resistors are constant. In practice, those resistors chane with respect to the power as well described in [7]. This is also the reason why there is no lon transitory in siulation, as explained in the experiental part. In view of these results, another advantae of the proposed current sharin can be underlined. As shown in Fi. 9, the proposed repartition leads to a lower current in the less efficient converter. Then, this converter will suffer lower stress than better converters. Finally, it can be assued that the proposed current-sharin schee leads to ake unifor aein of the paralleled odules. Only aein tests can corroborate this assertion but this aspect will not be treated in this paper. This property should have an effect on the structure health iprovin its lon-ter reliability and H. SHRAVANI, K. HTASWI facilitatin its aintenance. For exaple, a three-parallel boost converter structure can be easily realized usin interated circuits planned for inverters. In this case, the aein unifority resultin fro the proposed current-sharin schee will ensure a replaceent of the used odule when all the seiconductors are deficient. On the other hand, for classical ethods, the replaceent can be needed for only one of the converters presentin alfunctionin and wavefors as shown in Fis. to 8. Dis crete, Ts = 5e-06 s. powerui rs rs2 rs3 U3 Fi.9.Siulink desin of 3 parallel converters. 2 ref Mean (discrete) ref qu 8 qu 8 ref [Iin] Goto ref equ 0 U L Fro U2 L2 Fro2 U3 L3 Fro4 ' iload U U2 Fi.0. ontrol structure of converter. Fi..Input currents of converters. ' iload U_ U2_ U3_ Pout ref equ 9 Fro U_ Fro3 U2_ Fro5 U3_ Pout ref equ 20 o Pin ref Rp Pin ref equ 3 i ref i Display 00 ref i ref i equ 4 S Scope ref ref ref ontrol ontrol 2 ontrol 3 S i i ref equ 8 d eq 2 Po [Iin] Fro6 3 ik Product of leents Product Mean (discrete) d Divide Scope c PWM Generator -K- NOT Gai n ff Display U Goto U_ Goto

5 Fuzzy Induced fficiency Optiization of Paralleled D-D Boost onverter Fi.2. fficiency of converter. Fi.6. Input currents of converter with fuzzy. Fi. 3. Output voltae of converter. Fi. 7. fficiency of the converter with fuzzy. Fi. 4. ontrol structure of converter with fuzzy. Fi. 8. Output voltae of the converter with fuzzy. V. ONLUSION With the above results and raphs of the output voltae and efficiency coparison the converter has hiher efficiency as copared to control structure without fuzzy controller. The efficiency and voltae output of the conventional control converter is noted as 8% and 95V respectively. The efficiency and voltae output of the proposed fuzzy controlled converter is noted as 89% and 04V. However a sall ripple in the current is introduced which can be nelected for the increase in the efficiency of the converter. Fi.5. Fuzzy Interface Structure. VI. RFRNS [] A. Altowati, K. Zener, and H. Koivo, Modelin and control desin of paralleled D D switchin converters, presented at the Proc. Int. onf. oun., oput. Power, Sultan Qaboos University, Oan, 2009.

6 H. SHRAVANI, K. HTASWI [2] A. Izadian and P. Khayyer, Application of Kalan filters in odel-based fault dianosis of a D D boost converter, in Proc. I-36th Annu. onf. I Ind. lectron. Soc., 200, pp [3] G. Buiatti and A. Aaral, Paraeter estiation of a dc/dc buck converter usin a continuous tie odel, in Proc. ur. onf. Power lectron. Appl., 2007, pp. 8. [4] G. Buiatti, A. Aaral, and A. ardoso, An online technique for estiatin the paraeters of passive coponents in non-isolated D/D converters, in Proc. I Int. Syp. Ind. lectron., 2007, pp [5] B. Wan, arly oscillation detection for dc/dc converter fault dianosis, presented at the AIAA 9th Int. nery onvers. n. onf., San Dieo, A, USA, 20. [6] Y.-T. han and Y.-S. Lai, Paraeter tunin ethod for diital power converter with predictive current-ode control, I Trans. Power lectron., vol. 24, no. 2, pp , Dec [7] A. Shahin, A. Payan, J.-P. Martin, S. Pierfederici, and F. Meibody-Tabar, Approxiate novel loss forulae estiation for optiization of power controller of D D converter, in Proc. 36th Annu. onf. I Ind. lectron. Soc., 200, pp [8] J. Abu-Qahouq, Analysis and desin of N-phase currentsharin auto-tunin controller, I Trans. Power lectron., vol. 25, no. 6, pp , Jun [9] W. hen, X. Ruan, H. Yan, and. Tse, D D conversion systes consistin of ultiple converter odules: Stability, control, and experiental verifications, I Trans. Power lectron., vol. 24, no. 6, pp , Jun [0] P. Li and B. Lahan, A desin ethod for parallelin current ode controlled dc-dc converters, I Trans. Power lectron., vol. 9, no. 3, pp , May [] I. Kondratiev,. Santi, and R. Doual, Robust nonlinear syneretic control for -parallel-connected D D boost converters, in Proc. I Power lectron. Spec. onf., 2008, pp [2] R. Ahadi, H. Zararzadek, and M. Ferdowsi, Nonlinear power sharin controller for a double-input h- bride-based buckboost-buckboost converter, I Trans. Power lectron., vol. 28, no. 5, pp , May 203.