Torque Control Strategy for Parallel Hybrid Electric Vehicles using Fuzzy Logic

Transcription

1 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi orque Cotrol Strategy for Parallel Hybrid Electric Vehicles usig Fuzzy Logic Zhag Yi, Liu Hepig, Wag Huabi State Key Laboratory of Power rasmissio Equipmet & Syst Security ad New echology Chogqig Uiversity 29- Room of 6th Buildig of Departmet of Electrical Egieerig, 43 Chia zagyi8@yahoo.com.c Abstract: - his paper presets a desig method of a torque cotrol strategy for parallel hybrid electric vehicles (PHEV) by usig fuzzy logic. akig the driver commad, the state of charge (SOC) of the battery, ad the motor/geerator speed as iputs, a fuzzy cotroller ad relevat fuzzy rules have bee developed to effectively determie the torque distributio betwee electric motor (EM) ad iteral combustio egie (ICE). he uderlyig the of this strategy is to optimize the operatioal efficiecy of all compoets, cosidered as oe syst. he simulatio results reveal that, compared with the covetioal logic torque cotrol strategy (LCS) which uses precise threshold parameters, the proposed fuzzy torque cotrol strategy (FCS) improves fuel ecoomy ad maitais battery SOC withi its operatio rage more effectively. Key-Words: - Parallel hybrid electric vehicle; Hybrid powertrai; Cotrol strategy; Fuzzy cotrol; orque distributio Itroductio I recet years fuel ecoomy has bee oe of the domiate issues i automobile performace. Achievig the lowest possible fuel cosumptio cotributes to save atural resources ad traslates directly ito lower issios, which is ofte i cotrast with customers uiret i icreasig comfort ad performace. his has forced the auto idustry to search for ew types of vehicle that are more efficiet compared with covetioal vehicles. he potetial of so-called zero-issio vehicles iclude: fuel cell ad pure electric vehicle, is limited by fuel cell ad battery techology. he hybrid electric vehicle (HEV) which syergizes the electric power with the diesel egie has proved that its overall efficiecy is as high as a fuel cell from fossil fuel []. hus the HEV techology has bee proposed as the techology for ext geeratio vehicle cofiguratios, ad the rule of cotrol strategy i the hybrid drive-trai is escalatig. A maaget or cotrol strategy, which is usually impleted i the vehicle cetral cotroller, is defied as a algorithm, or a law regulatig the operatio of the drive-trai of the vehicle. Geerally, it iputs the measurets of the vehicle operatio coditios such as speed or acceleratio uested torque by the driver. he outputs of a cotrol strategy are decisios to tur o or off certai compoets or to modify their operatig regios by commadig local compoet cotrollers [2][3][4]. he cotrol or maaget strategies for parallel HEVs ca be roughly classified ito two categories. he first approach is based o static optimizatio methods. Commoly, electric power is traslated ito a equivalet amout of (steadystate) fuel rate i order to calculate the overall fuel cost [5][6]. he optimizatio sche the figures out the proper split betwee the two eergy sources (ormally the iteral combustio egie ad the electric motor) usig steady-state efficiecy maps. But it s very difficult to determie the exact chageover poit i the costructio of a cotrol strategy. Furthermore, may trial-ad-error-based tests modifyig the drivig strategy should be uired to get the effective oe. o solve this issue, the secod type ploys heuristic cotrol techiques such as cotrol rules/fuzzy logic/eural etworks for estimatio ad cotrol algorithm developmet [7][8][9], which has a meas of cosiderig the dyamic ature of the syst whe performig the optimizatio. Regardig a hybrid drive-trai as a multidomai, oliear ad time-varyig plat, fuzzy logic, the usefuless of which for decisio makig for a ucertai ad imprecise plat has already bee itroduced i may idustrial fields, ses to be the most logical ad feasible approach to the ISSN: Issue 4, Volume, April 2

2 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi Fig. Parallel HEV cofiguratio probl []. I fact, istead of usig determiistic rules, the decisio-makig property of fuzzy logic ca be adopted for realizig a real-time ad suboptimal torque split. I aother words, fuzzy logic cotroller is a extesio of the covetioal rule based cotroller. I this paper, a torque cotrol strategy based o fuzzy logic for parallel hybrid electric vehicles is developed to optimize the operatio of all major PHEV compoets. he orgaizatio of the paper is as follows. Firstly, itroduces the PHEV cofiguratio ad briefly describes the simulatio model. Subsequetly, the torque maaget strategy is preseted. I additio, it followed by the descriptio of the fuzzy logic power cotroller. Fially, the simulatio results are preseted to ivestigate the effectiveess of the fuzzy torque cotrol strategy. 2 PHEV Cofiguratios here are geerally two accepted basic cofiguratios for the hybrid electric vehicles icludig series ad parallel. he multi-mode ad complex type which is also kow as series-parallel type is cosidered combies the features of both the series ad parallel hybrids []. I series HEVs, a geerator coverts the ICE mechaical output ito electricity which either charges the battery or ca bypass the battery to propel the wheels via the same electric motor ad mechaical trasmissio. Beefited from the decouplig betwee the egie ad the drivig wheels, there is a advatage of flexibility for locatig the ICE geerator set. Despite its advatage of simplicity of the drivetrai, it eeds three propulsio devs, the ICE, the geerator, ad the electric motor, which defiitely damages the efficiecy of SHEV. Aother disadvatage is that all these propulsio devs eed to be sized for the maximum sustaied power if the SHEV is desiged to climb a log grade. Fig presets the block diagram of parallel HEVs with the ICE ad EM. Differig from the SHEV, the PHEV allows both the ICE ad EM to deliver torque i parallel to drive the wheels. S both of th are geerally coupled to the drive shaft of the wheels via two mechaical trasmissios, the propulsio torque may be supplied by the ICE aloe, by the EM aloe or by both. he EM ca be used as a geerator to charge the battery by regeerative brakig or absorbig torque from the ICE whe the output torque is greater tha that uired to drive the wheels. Better tha the series HEV, the parallel hybrid uses the ICE to drive the vehicle shaft directly, which avoids the losses i power coversios. Aother advatage of the series case is that a smaller ICE ad a smaller EM ca be used to get the same performace util the battery is depleted. Eve for log trip operatio such as urba ad expressway workig coditio, oly the ICE eeds to be rated for the maximum sustaied power, while the EM may still be about half. I brief, parallel HEV gets a loger drivig rage tha series HEV at the cost of complicated syst architecture, high uiret of cotrol syst ad the cosequet high pr. 2. orque Distributio o PHVE he differece of torque path betwee the ICE ad the EM is explaied i Fig 2. Whe the ICE is used, the torque drives directly from the ICE to the wheels (path ). Whe the EM is used as a motor, the EM output positive torque to drive the wheels (path 2). Whe the EM is used as a geerator, the EM absorbs torque to charge the battery (path 3). Fig 2. Block diagram of the torque path ISSN: Issue 4, Volume, April 2

3 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi Depedig o the torque distributio, there are five differet ways to operate the parallel HEV syst: ) Provide torque to drive the wheels with oly the ICE (path ); 2) Oly the EM (path 2); 3) Both the ICE ad the EM simultaeously (path ad 2); 4) Charge the battery, usig part of the ICE torque (path ad 3); 5) Regeerative break by drivig the EM as a geerator that provides power to the battery (path 3). So a torque cotroller is eeded to maage the torque distributio betwee the ICE ad EM. he cotroller adds the capability for the compoets to work together i harmoy, while at the same time optimizes the operatig poits of the idividual compoets. 2.2 Syst Cofiguratio 4) Electroic cotrol automatic trasmissio (ECA): five speed; 5) otal test vehicle mass: 5kg. he details of the powertrai syst for CQU- PHEV are give i Fig 3. Accordig to the Natioal Stadard of the People s Republic of Chia: Hybrid electric vehicles Power performace est method (GB/ ), this PHEV is desiged to fulfill the performace uirets described i able. 3 orque Cotrol Strategy his sectio describes the torque cotrol strategy, which is the philosophy behid the torque cotroller. he torque distributio i the syst should be maaged i such three rules that: ) he SOC of battery should ever drop too low; 2) he driver iputs (form acceleratig ad brake pedals) are satisfied cosistetly except where it would coflict with the first rule; 3) he overall efficiecy of the mai compoets (ICE, EM, battery, ad trasmissio) is optimized as log as such optimizatio does t coflict with first or secod rule. Fig 3. Block diagram of the powertrai for a PHEV able. Vehicle performace uirets stated by GB/ Maximum speed km/h Acceleratio ability ~5km/h 5sec Acceleratio ability ~km/h 3sec Speed uphill ability i road slope of 4% 8km/h Speed uphill ability i road slope of 2% 65km/h Hill startig ability 3% Uphill ability o maximum road slope 3% he specific PHEV cofiguratio, used throughout the paper, cosists of the followig compoets: ).4L ilie four cyliders ad electroic ijectio gasolie egie: 5kW; 2) AC asychroous motor: 2kW cotiuous, 4kW peak; 3) Lithium iro phosphate battery: 3V DC, Ah; 3. Fudametal Operatio Modes While the PHEV is operated, the torque cotrol strategy is uired to decide how much torque is eeded to drive the vehicle ad how much is eeded to charge the battery. he the torque uiret is split betwee the ICE ad EM. his torque-split strategy optimizes the efficiecy of all mai compoets of the PHEV, because it determies the operatig poits of the compoets. o idetify the optimal operatig poits, the operatio modes of vehicle ad the correspodig efficiecy maps of compoets were studied. he Fig 4 illustrates the fudametal operatio modes. I particular, these five parts of the curved lies are tied to the appropriate operatio modes ad form a itegral part of the drivig cycle as follows: ) he curve segmet AB is the stadig start acceleratio coditio related to the EM separate drivig mode; 2) the curve segmet BC is the ormal acceleratio coditio related to the ICE separate drivig mode; 3) the curve segmet CD is the cruisig coditio related to the ICE drivig ad battery chargig mode (also kow as active chargig mode); ISSN: Issue 4, Volume, April 2

4 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi 4)the curve segmet DE is the sap acceleratio coditio related to the ICE ad EM parallel cooperatig mode (also kow as EM assistace mode); 5) the curve segmet EF is the breakig coditio related to EM regeerate breakig mode opt off Fig 5. he ICE efficiecy map ad the rule based cotrol strategy Fig 4. he typical drivig cycle ad the fudametal operatio modes 3.2 orque Split Strategy Based o the available output torque, the pedal positio is chaged ito the uirig torque. If <, it s regeerate breakig, of which the cotrol strategy is to recover as much eergy as possible uder the limit of the EM ad battery, meawhile the frictioal dampig dev is supposed to meet the shortage of the breakig torque. If >, the uirig torque is split betwee the EM ad ICE: = + () is the output torque of ICE. is the output torque of EM which could be positive i EM assistace mode, ad egative i EM geeratio mode. Accordig to the uiversal characteristic curve of ICE i Fig 5, is adjusted to optimize the ICE load. here are ICE optimal efficiecy curve ( ) which is used as the optimal opt objective of the ICE operatig poits, maximal torque curve ( ) ad ICE turoff curve ( off ) i Fig 5. Whe < off, the ICE is turoff to void workig at ieffective coditio. he ICE drives the EM to charge the battery which calls active charge (show i Fig 5); while i regeerative brakig coditio, it s passive charge. he ICE load is adjusted to maitai it i high effective workig coditio by the meas of active charge or EM assistace. It s ecessary for cotrol strategy to maitai the battery SOC i workig sectio i which the battery charge ad discharge with low iteral resistace. he lowest poit of battery iteral resistace is defied to target SOC i Fig 6. If the SOC is lower tha it, the possibility of active charge will icrease, while o the cotrary it s more iclied to discharge (EM assistace or drivig aloe). Resistace / Ω Fig 6. Charge ad discharge resistaces of the battery off could be defied by comparig the eergy efficiecy η i EM drivig mode with η 2 i ICE drivig mode as follow: η = η η η η η (2) bat tra η = η η (3) 2 tra η is the ICE efficiecy, η is the EM efficiecy, η is the mechaical efficiecy of tra ISSN: Issue 4, Volume, April 2

5 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi q SOC λ Fig 7. Block diagram of fuzzy torque cotroller trasmissio, η bat is battery charge-discharge efficiecy. η i (2) is quite higher tha the oe i (3) owig to ICE is more effective i chargig battery tha i drivig vehicle. Whe η2 η, the ICE works at low effective coditio. For istace, if the ICE, EM ad battery efficiecy i (2) is.3,.85 ad.9, the ICE turoff threshold is set to.2 that meas ICE should be turoff if its efficiecy was below.2. Ad the correspodig ICE torque is defied as off. But whe referrig to practical circumstaces, it could be much complicated. I active chargig mode, the fuel cosumptio icreased by extra load is limited as a result of improvet i ICE efficiecy. Furthermore, there is o fuel cost i passive chargig mode. Accordigly, actual off should be a little higher tha theoretical value. I the view of the difficulty i real-time quatitative aalysis ad comparig eergy coversio efficiecy, it s a practical method to predefie a set of threshold values for prescribig ad ICE high off efficiecy area which is aroud opt. he gear shiftig cotrol strategy divides ito two broad categories: whe ICE o, the ICE efficiecy is the prime cosideratio to choose the optimal shift; Otherwise (EM separate drivig mode), the EM efficiecy gets the priority. his strategy will be discussed i more detail at a later chapter. With the help of logic threshold, the cotrol strategy metioed above is coverted ito a set of If-the rules to cotrol the output torque distributio ad switchover betwee the workig modes, also kow as logic torque cotrol strategy (LCS). he ext sectio discusses how the driver torque commad is computed ad how fuzzy logic is applied to compute the optimal ICE torque. 4 Fuzzy Logic orque Cotroller O the grouds of LCS described i the previous sectio, a fuzzy torque cotrol strategy (FCS) is built by replacig the boolea logic ad accurate parameters with the fuzzy logic ad fuzzy parameters [2][3]. A block diagram of the FCS cotroller is preseted i Fig 7, ad the fuzzy logic cotroller (FLC) beig the mai part. he basic idea of a FLC is to formulate huma kowledge ad reasoig, which ca be represeted as a collectio of if-the rules, i a way tractable for computers. 4. Driver Commad Iterpreter he first block of fuzzy torque cotroller (FC) is Driver Commad Iterpreter (DCI) which coverts the driver iputs from the brake ad accelerator pedals to a driver torque uirig. he sigals from the pedals are ormalized to a value betwee zero ad oe (Oe: the fully pressig positio, zero: the iitial positio). he breakig pedal sigal is subtracted from the acceleratig pedal, so that the driver iput takes a value betwee - ad +. Whe the driver iput below zero, it is sed to a separate brake cotroller (ot show i Fig 7) which computes the regeerative brakig ad the frictio brakig torque uired to decelerate the vehicle. Ad the driver uirig torque seds to FLC is forced to zero. O the cotrary, the positive driver iput is multiplied by at curret vehicle speed to get max. he ICE speed ad tperature perform collaboratively o which is computed usig a tow-dimesioal look-up table with speed ad tperature as i puts. However, for a give vehicle speed, the ICE speed has oe out of five possible values (oe for each gear umber of the trasmissio). At first, the maximum ICE torque levels for those five speeds are computed. he the maximum of these values is selected as, max ISSN: Issue 4, Volume, April 2

6 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi while is divided by the correspodig opt ad the quotiet q (betwee ad 2) is set to Fuzzy Logic Cotroller (FLC). here is oe exceptio to the gear shiftig cotrol strategy: whe vehicle works at EM separate drivig mode, the gear umber is chose to maximize the EM output torque istead of the ICE. 4.2 Fuzzy Logic Cotroller Oce the driver torque commad q is arrived, the secod block of fuzzy torque cotroller, amely FLC, calculates the optimal EM torque i the form of ormalized value ( λ ) betwee -.5 ad.5. Because the EM speed is equal to the ICE speed, is selected alog with. herefore, the EM objective torque could be defied as follow: = λ (4) obj max 4.2. Mbership Fuctios he crisp iputs are coverted to fuzzy i Fuzzificatio process by usig trapezoid mbership fuctios which are described by the uiform equatio. µ ( x) = A, x< mi A or x> max A (5) y, y < ad KA, ( y2 > or K 2A = ) ad ( y > or K ) A = y2, ( y2 < ad K 2A ) ad ( y > or K = A ) y = ( x mi ) K (6) A A y2 = ( x ) K 2 (7) max ad mi A A are the upper ad lower limits respectively of oe fuzzy set. K A ad K 2A are the slopes of lie ad lie 2 respectively to the x axis i Fig 8. If lie or lie 2 is vertival to the x axis, K A or K 2A is zero. µ ( x) Al A 2 KA K 2A A Am A mi x A max A Fig. 8 rapezoid mbership fuctio As the x is withi mi A ad max A, this equatio firstly calculates the degree y of mbership by the lie usig the liear equatio of lie. If y is t greater tha, the degree of µ A is equal to y ; otherwise it will be recalculated by the liear equatio of lie 2. If y ad y 2 are both greater tha, µ A is equal to. he fuzzy set will be easy to adjust by calibratig the umerical array { mi, max, A A K A, K 2A }. I this paper, the two iputs of the FLC are: q ad SOC, while the output is: λ. Fig 9 presets the mbership fuctios (MFs) for q, SOC ad λ. µ ( q ) µ ( SOC ) ( ) µ λ LL L M H HH q LL L M H HH SOC NL NM NS Z PS PM λ Fig. 9 he mbership fuctios for the iputs ad output of the FLC Fuzzy Rules he rule base is preseted i able 2. Accordig to previous study [4], the fuzzy rules are divided ito six parts which are as follow: I. If the SOC is low, the battery is supposed to be charged to a higher level. Especially whe the SOC drops to very low, chargig the battery should be a top priority. his will result i a relatively fast retur of the SOC to moderate level. II. If drops to very low level, i aother word, q is very low, the ICE works quite iefficiet. It should be tured off usig the EM to supply aloe (EM separate drivig mode). III. If is slightly lower tha opt, the EM work as a geerator (EM active chargig mode) to add extra load. So that the ICE workig poit could ISSN: Issue 4, Volume, April 2

7 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi be raised to approach the ICE optimal efficiecy curve. IV. If is close to opt, the ICE works quite effectively already. No heavy extra load should be added i (ICE separate drivig mode), or else it could damage the efficiecy. V. If is slightly higher tha opt, it s similar to III, except the EM output torque is positive (EM assistace mode). VI. If rises to a very high level, all torque should be available to maitai the uiret, regardless of the efficiecy. able 2. Fuzzy rules for fuzzy torque cotroller λ q LL L M H HH SOC LL L M H HH NL NL NL NM NS PS NM NS Z NM Z PS PS NM PM PM NS PM Fuzzy Iferece Assume the total umber of the fuzzy rules is N ad A, B, C deote respectively the iput value: q, SOC ad output value: λ of the FLC i rule ( =, 2,..., N). he fuzzy reasoig mechaism used i this paper ca be separated i five mai steps. ) Fuzzificatio: he mbership degrees of the two iputs of the FLC are computed via the mbership fuctios. µ ( q), µ ( SOC) (8) A 2) Rule stregth: Accordig to the fuzzy rules, the fuzzified iputs are combied to establish the rule stregth ω which is the degree to which atecedet part of a fuzzy rule is satisfied ad also kow as the degree of fulfillmet. ω = mi( µ ( q), µ ( SOC)) (9) B A B 3) Iferece: his operatio represets the ifthe implicatio. he rule stregth of the atecedet of each rule is used to modify the cosequet of that rule accordigly. his is doe by multiplyig the degree of fulfillmet of the atecedet ( ω ) with the cosequet ( µ C ( λ ) ) of rule. µ ( λ) = ω µ ( λ) () CO C 4) Aggregatio: he outputs of all of the fuzzy rules are combied to obtai oe fuzzy output distributio. his is doe by takig the summatio of the iferece results of each rule. N µ ( λ ) = µ ( λ) () C = CO 5) Defuzzificatio: his operatio takes the output distributio ad fids its ceter of mass to come up with oe crisp umber (λ ). w is the mass cetre of µ ( λ ). CO λµ C ( λ) dλ λ= = µ ( λ) dλ C N w µ C ( ) O λ dλ = N = µ ( λ) dλ CO (2) 4.3 ICE ad EM orque Adjustor he third block i Fig 7 computes the fial values for the ICE torque ( ) ad EM torque ( ), usig, obj, off. Normally, equals to obj, ad is. But there are two exceptios: ) Whe obj > ad obj is smaller tha off, the = ad =. his meas the vehicle works at EM separate drivig mode, ad the gear shift should be recomputed to make the maximum EM torque possible. 2) Whe obj is larger tha which is selected i the first block of FC, the = ad = obj. 5 Simulatio Results he proposed FCS has bee impleted ad simulated by the advaced vehicle simulator (ADVISOR) i MALAB/SIMULINK eviromet. ADVISOR ploys a combied forward/backward facig approach for the vehicle performace simulatio. he simulatio parameters as well as the vehicle compoets model are cosistet with the cofiguratio i chapter 2. Ad its simulatio results have bee compared with the LCS s, so as to get a better view of the advatage ad improvets i FCS. Fig presets the compariso of the torque distributio i FCS ad LCS uder the New Europea Drivig Cycles (NEDC) with the iitial SOC of.65. i FCS is more homogeeous tha the oe i LCS. Especially uder urba drivig cycle (s 8s), i LCS stays at a ISSN: Issue 4, Volume, April 2

8 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi / ( ) v km i h / ( ) v km i h SOC.6 SOC.6 / ( ) Ni m / ( Ni m ) / ( Ni m ) t / ( s) (a) FCS / ( ) Ni m / ( Ni m ) / ( Ni m ) t / ( s) (b) LCS Fig. Simulatio results for FCS ad LCS uder NEDC Fig. orque distributio by FCS i differet workig mode low level, which idicates the ICE work at low effective zoe i low speed drivig coditio. I order for full uderstadig of the torque distributio disciplie i FCS, Fig illustrates the details i Fig (a) from 98s to 85s. ) From 987s to 36s, maitais at a low level ad the SOC rais moderate, so the EM provides all of the drivig force. 2) From 37s to s, icreases ad the ICE starts up, followig the vehicle speeds up. But is isufficiet to maitai the ICE works at high effective zoe, so the EM performs active charge to raise the ICE load. 3) From 2s to 32s, as the vehicle speed cotiues to icrease, climbs up to approach. he EM should provide extra torque to assistat the ICE to meet the uiret. 4) From 33s to 67s, as the vehicle begis to parkig brake, the EM charges the battery agai ad outputs as much egative torque as possible to esure always maximizes the regeerative brakig power. Fig 2 presets the ICE operatig poits o efficiecy map of the simulatio experimet illustrated i Fig. he operatig poits i Fig 2 (a) are close to the optimal curve, which idicates that the ICE has bee operated close to optimal efficiecy. By compariso, the operatig poits i Fig 2 (b) are farther from the optimal curve tha those for the FCS. What is worse is that most of the operatig poits are spread out all over the map rather tha stick with the optimal curve i the low speed zoe of Fig 2 (b). his echoes the coclusio proposed before that the ICE efficiecy for FCS is defiitively higher tha the oe for LCS i low speed drivig coditio, which could reflect o a sigificat gap i the ICE efficiecy. ISSN: Issue 4, Volume, April 2

9 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi Fig 2. ICE operatig poits for FCS ad LCS o efficiecy map able 3. he mai compoets efficiecy ad fuel ecoomies uder NEDC revised η η η bat η tra F ( L / km) he efficiecy of the mai compoets uder NEDC is show i able 3. S the EM ad battery s high effective zoe is relatively widespread ad their efficiecy is isesitivity to drivig load, the efficiecy variace i EM ad battery betwee FCS ad LCS is t as rarkable as it for ICE. However, there is still some advatage for FCS i EM ad battery, ad therefore the FCS overall efficiecy is better tha the LCS. o be able to compare the vehicle fuel ecoomy of FCS ad LCS, the battery eergy at the begiig ad the ed of each cycle has to be the same. herefore, a series of test with SOC form.7 to.35 is carried out. hrough liear regressio aalysis of the relatioship betwee the fuel cosumptio ad SOC chages, we obtai the fuel cosumptio with SOC uchaged ( F revised ) for FCS ad LCS, respectively. he overall improvet of FCS tha LCS equals 9.4% which explifies the advatage of Fuzzy cotrol strategy over covetioal cotrol strategy. 6 Coclusios I this paper, a fuzzy torque cotrol strategy for parallel hybrids has bee preseted. o implet this strategy, the accelerator ad brake pedal iputs of the driver are coverted to a driver uirig torque at first. Ad the the uirig torque is divided by the ICE optimal torque which is selected i gear shiftig cotrol strategy with the ICE maximum torque. heir quotiet ad state of charge of the battery are used by a fuzzy logic cotroller to compute the ormalized value of EM objective torque. I the ed, the EM objective torque, driver uirig torque, ad ICE off torque are used to compute the ICE torque ad EM torque which is sed to ICE cotroller ad EM cotroller, respectively. he torque cotrol strategy trades off betwee the ICE efficiecy ad the efficiecy of the other compoets, meawhile esures that the battery is sufficietly charged at all times, the driver iputs (form brake ad accelerator pedals) are satisfied cosistetly, ad the fuel ecoomy of the PHEV is optimized. he simulatio results uder NEDC show potetial improvet by applyig fuzzy logic, over covetioal strategy that relies o Boolea logic. I further research, the robustess of the fuzzy logic cotroller will be ivestigated i more detail. Ad eural etwork or geetic algorithms will be added to cotroller, to eable o-lie cotroller optimizatio. Refereces: [] S. S. Williamso, A. Emadi, Comparative assessmet of hybrid electric ad fuel cell vehicles based o comprehesive well-towheels efficiecy aalysis. IEEE rasactios o Vehicular echol, vol.54, o.3, 25, pp ISSN: Issue 4, Volume, April 2

10 WSEAS RANSACIONS o SYSEMS Zhag Yi, Liu Hepig, Wag Huabi [2] C. C. Cha, he state of the art of electric, hybrid, ad fuel cell vehicles, Proc. of the IEEE, vol.95, 27, pp [3] F. R. Salmasi, Desigig cotrol strategies for hybrid electric vehicles, utorial Presetatio i EuroPes 25, Jue 5-7, Bealmadea, Spai, 25. [4] L. Gheorghe, H. Vasile, A. Mihai, R. Marcel, Evaluatio of cotrol algorithms for hybrid electric vehicles, WSEAS rasactios o Systs, vol.6, 27, pp [5] P. Caratozzolo, M. Serra, J. Riera, Eergy maage strategies for hybrid electric vehicles, Proceedigs of the IEEE Electric Machies ad Drives Coferece (IEMDC 3), 23, pp [6] C. C. Li, Huei Peg, J. W. Grizzle, Power maaget strategy for a parallel hybrid electric truck, IEEE rasactios o Cotrol Systs echology, vol., o.6, Nov. 23, pp [7] H Zhog, F Wag, G-Q Ao, J-X Qiag, L Yag, B Zhuo, X-J Mao, A optimal torque distributio strategy for a itegrated startergeerator parallel hybrid electric vehicle based o fuzzy logic cotrol, Proc. IMechE J. Automobile Egieerig, vol.222, part D, Sep. 27, pp [8] N. A. Kheir, M. A. Salma, N. J. Schoute, Emissios ad fuel ecoomy trade-off for hybrid vehicles usig fuzzy logic, Math. Computer Simulatio, vol.66, 24, pp [9] N. J. Schoute, M. A. Salma, N. A. Kheir, Eergy Maaget strategies for parallel hybrid vehicles usig fuzzy logic, Cotrol Egg Pract, vol., 23, pp [] B. M. Bauma, G. Washigto, C. G. Bradley, ad R. Giorgio, Mechatroic desig ad cotrol of hybrid electric vehicles, IEEE/ASME rasactios o Mechatroics, vol.5, o., Mar 2, pp [] Y. Gao, M. Ehsai, J. M. Miller, Hybrid electric vehicle: overview ad state of the art, Proceedigs of the IEEE Iteratioal Symposium o Idustrial Electroics 25, vol., 25, pp [2] C. C. Lee, Fuzzy logic i cotrol systs: fuzzy logic cotroller. I, IEEE rasactios o Systs, Ma ad Cyberetics, vol.2, 99, pp [3] V. Costati, Properties of fuzzy systs, WSEAS rasactios o Systs, vol. 8, 29, pp [4] D.. Qi, M. H. Hu, Y. L. Yag, H. Shu, Optimizatio of Syst Efficiecy for the Mild Hybrid Electric Vehicle with Cotiuously Variable rasmissio uder the Motor ad Egie Combied Workig Coditios, Chiese Joural of Mechaical Egieerig, vol., o., 28, pp ISSN: Issue 4, Volume, April 2