OptimizationofHybridEnergyStorageSystemControl...

Research ArticleOptimization of Hybrid Energy Storage System ControlStrategy for Pure Electric Vehicle Based on Typical Driving Cycle

Kanglong Ye1 Peiqing Li 123 and Hao Li1

1School of Mechanical and Energy Engineering Zhejiang University of Science and Technology Hangzhou 310023 China2School of Mechanical Engineering Zhejiang University Hangzhou 310058 China3Zhejiang Fuchunjiang Communication Group Hangzhou 310089 China

Correspondence should be addressed to Peiqing Li lpqinghotmailcom

Received 31 January 2020 Revised 24 May 2020 Accepted 27 May 2020 Published 29 June 2020

Academic Editor Haopeng Zhang

Copyright copy 2020 Kanglong Ye et al is is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Taking a hybrid energy storage system (HESS) composed of a battery and an ultracapacitor as the study object this paper studiesthe energy management strategy (EMS) and optimization method of the hybrid energy storage system in the energy managementand control strategy of a pure electric vehicle (EV) for typical driving cycles e structure and component model of the HESS areconstructed According to the fuzzy control strategy aimed at the roughness of the membership function in EMS optimizationstrategies based on a genetic algorithm (GA) and particle swarm optimization (PSO) are proposed these use energy consumptionas their optimal objective function Based on the improved EV model the fuzzy control strategy is studied in MATLABAdvisorand two control strategies are obtained Compared with the simulation results based on three driving cycles urban dynamometerdriving schedule (UDDS) new European driving cycle (NEDC) and ChinaCity the optimum control strategy were obtainedetheoretical minimum energy consumption of HESS was reached by dynamic programming (DP) algorithm in the same simulationenvironment e research shows that compared with the PSO the output current peak and current fluctuation of the batteryoptimized by the GA are lower and more stable and the total energy consumption is reduced by 3ndash9 in various simulation casestudies Compared with the theoretical minimum value the deviation of energy consumption simulated by GA-Fuzzy Controlis 06

1 Introduction

In recent years new energy vehicles have become the maindevelopment direction of the automobile industry Com-pared with fuel vehicles pure electric vehicles have thecharacteristics of energy saving and environmental pro-tection without exhaust pollution However owing to theinfluence of the battery material power supply form powermanagement strategy and driving environment pureelectric vehicles (EVs) have defects such as short endurancemileage and poor energy consumption stability Amongthem whether the energy management and control systemof a hybrid energy storage system (HESS) can provide thebest energy distribution strategy according to changes in theworking conditions has become one of the hot spots incurrent research on the energy management and control of

EVs By improving the power management and controlsystem of electric vehicles the stability of the managementstrategy and endurance mileage of EVs can be improved

To improve the energy control efficiency of EVs andemphasize real-time optimization domestic and overseasscholars have conducted significant amounts of research onenergy control strategies and predictive control Somescholars optimized the working efficiency of the powersystem by improving the components of the HESS In [1 2]a new hybrid batteryultracapacitor energy storage systemfor electric vehicles (including electric vehicles hybrid ve-hicles and plug-in hybrid vehicles) was proposed issystem uses a smaller DCDC converter as a controlledenergy pump to keep the voltage of the ultracapacitor higherthan that of the battery under urban driving conditions In[3] a modulator replaces the DCDC converter for

HindawiMathematical Problems in EngineeringVolume 2020 Article ID 1365195 12 pageshttpsdoiorg10115520201365195

connections which solves the problem of large voltagechanges caused by ultracapacitor power transmission usthe drive performance of the motor is not disturbed

Research on energy management strategies can optimizethe energy efficiency of the entire vehicle without changingthe basic components and framework of the HESS In [4] anoptimization framework was proposed to calculate thesuboptimal current of the hybrid system in an EV to controlthe current and minimize the working current and fluctu-ation of the battery in the EV In [5ndash8] by combining abattery and ultracapacitor an energy control strategy withfuzzy control strategy as the core was proposed to improvethe fuel economy and durability of energy system compo-nents while maintaining the vehicle power performance In[9 10] a real-time optimal energy management strategy(EMS) was proposed for a plug-in hybrid bus based on theminimum equivalent fuel consumption strategy and con-sidering the frequent starting of a low-speed engine In[11 12] the energy management of a hybrid power systemwas analyzed e energy flow of different energy sourceswas managed separately by combining fuzzy logic controland shape control In [13] the online predictive controlstrategy of a series and parallel plug-in hybrid EV wasstudied A new dual-loop online intelligent planning (DOIP)method for speed prediction and energy flow control wasproposed and a depth fuzzy predictor was established torealize directional speed prediction e optimal controlbehavior was determined by learning the vehicle speed andacceleration

ere are also some scholars who use algorithms tooptimize research Reference [14] has proposed an adaptivepower distribution algorithm e parameters of the algo-rithm were optimized by combining self-organizing map-ping and particle swarm optimization (PSO) to alleviate thepeak demand and short charge and discharge period of thebattery In [15 16] the focus is on the k-means clusteringalgorithm Reference [15] focuses on the study of the drivingcycle of cars in Tehran and its suburbs By collecting thesituation of vehicles running in actual traffic the calculationis based on the definition of ldquomicro-travelrdquo Two drivingfunctions ldquoaverage speedrdquo and ldquoidle time percentagerdquo arecalculated e micro-trips are then clustered into fourgroups in driving feature space using the k-means clusteringmethod e focus of [16] is the application of drivingcondition recognition in hybrid electric vehicle intelligentcontrol For this purpose driving features are identified andused for driving segment clustering using the k-meansclustering algorithm Many combinations of driving featuresand different numbers of clusters are evaluated in order toachieve the best traffic condition recognition results Ref-erences [17 18] also record the vehicle data and driving dataunder the actual traffic conditions analyze the driving endand study the influence of driving characteristics on the fuelconsumption and exhaust emissions through the drivingsegment simulation e results of [17] show that the ve-locity-dependent driving features such as ldquoenergyrdquo ldquomean ofvelocityrdquo ldquodisplacementrdquo and ldquomaximum velocityrdquo aremore effective in vehicle exhaust emissions and fuel econ-omy e results of [18] show that ldquoenergyrdquo and ldquopercentage

of free timerdquo are two driving characteristics to drive seg-mented clustering Driving segment clustering can be usedin driving cycle development intelligent hybrid vehiclecontrol and so on

In the current study most of the control strategies of EVsare optimized in one algorithm Based on the above researchthis paper uses fuzzy control strategy as an EMS with thecomposite power supply form of a combined battery andultracapacitor Reducing the total energy consumption is theoptimization goal and GA and PSO are used to carry outlearning optimization for the fuzzy control strategy inMATLABAdvisor in the UDDS NEDC and ChinaCitydriving cycles In the same simulation environment thetheoretical minimum energy consumption of HESS is cal-culated by DP algorithm An experimental analysis is carriedout to compare the battery current output performance andthe total energy consumption parameters of the energysystem and to evaluate the optimization effect of thealgorithm

2 System Description and Methodology

21 Modeling of HESS ere are three common types ofcomposite power supply structure passive semiactive andactive e semiactive structure has two types battery endload and capacitor end load e main feature of the passivehybrid energy system is that the battery and the ultra-capacitor are directly connected in parallel on the power busand the voltage of the two is synchronized in real time suchthat the energy distribution between the battery andultracapacitor cannot be flexibly adjusted in the case of high-power output or full capacity e composite power supplywith an active structure has two DCDC converters that areconnected in series with batteries and capacitors e twoDCDC converters are integrated and connected in seriesBoth of these forms can flexibly adjust the distribution of theenergy output but the energy system structure is complexand the actual application cost is high [19]

erefore in this study the battery terminal loadstructure of a semiactive energy system is selected Its to-pology is shown in Figure 1 e battery and capacitor areisolated by a DCDC converter e battery acts as the mainpower supply and the capacitor is the auxiliary powersupply e energy system distribution of the battery andcapacitor can be realized through high and low voltageconversions of the voltage Compared with the battery endload the ultracapacitor is used as the main power supply inthe form of the capacitor end load Owing to the low capacityof the capacitor energy the battery needs to participate indriving and all current flowing through the battery needs topass through the DCDC converter is makes the batterywork with low efficiency and high energy consumption

22 Modeling of Battery Battery modeling is a significanttask within battery technology development and is vital inapplications For example EV range prediction is onlypossible through the application of advanced batterymodeling and estimation techniques to determine current

2 Mathematical Problems in Engineering

state and predict remaining endurance In addition batterymodeling is essential for safe charging and dischargingoptimal utilization of batteries fast charging and otherapplications [20]

A battery simulationmodel verifies the correctness of themodel parameter settings in a simulation Generally theinput is the current and the output is the terminal voltageBecause the current power temperature state of charge(SOC) and other parameters have nonlinear effects on thebattery characteristics considering all factors in modelingmakes the simulation calculation too large and difficult tocontrol

Equivalent circuit models of a battery include the Rintmodel the evenin model and the second-order reservecapacity (RC)modele Rint model is the equivalent circuitmodel of internal resistance which regards the battery as aseries model of the ideal voltage source and resistance Inthis model it is easy to set the parameters and run a sim-ulation but the accuracy is low e evenin model is afirst-order RC model and contains a voltage source and anRC parallel circuit e model fully considers the relation-ship between the electromotive force and SOC and thedynamic process of the battery It can accurately simulate thebattery charging and discharging process but it does notconsider the open-circuit voltage changes caused by thecurrent accumulation so it is not suitable for long-timesimulationse second-order RCmodel adds a group of RCcircuits on the basis of theeveninmodel In the model thevariable voltage source connects the resistance and two RCcircuits is can provide better consideration to the tran-sient and steady-state characteristics of the battery but doesnot consider the influence of the temperature and batteryself-discharge [21 22]

is paper compares the optimization effects of differentalgorithms and does not require a high-precision simulationso the more universal Rint model is selected for modelingFigure 2 shows the battery equivalent circuit model and themathematical model is described as follows

Qb nb1nb2Qbc

Rb nb1Rbc

nb2

Ub nb1Ubc

dSOCb

dt

Ub minusU2

b minus 4RbPm

1113969

2RbQb

Pb minusdSOCb

dtUbQb

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(1)

where Qb Rb and Ub respectively represent the capacityinternal resistance and terminal voltage of the battery packQbc Rbc and Ubc respectively represent the capacity in-ternal resistance and terminal voltage of the battery cell nb1and nb2 respectively represent the number of series andparallel modules in the battery pack and Pb is the powerdensity of the battery

23 Modeling of Ultracapacitor An ultracapacitor is a typeof electrochemical element that stores energy by virtue ofphysical characteristics Unlike a battery with a large ca-pacity the ultracapacitor has a higher energy density highercharge and discharge power and longer cycle life It issuitable to use for power transport in the start and stopstages active suspension systems and rapid accelerationstage In recent years ultracapacitors have been widely usedin high-power energy storage systems of vehicles ships andaerospace projects [23ndash25]

e RC internal-resistance model which is common andeasy to implement is also selected to describe the ultra-capacitor e model is generally composed of a series re-sistance parallel resistance and ideal capacitor eequivalent circuit model is shown in Figure 3

DCDC converter Ultracapacitor

Battery

Motor controller

Driving motorLoad

Figure 1 Model of the vehicle electrical energy system

I

U

+

ndash

Figure 2 Internal-resistance model of the battery

Mathematical Problems in Engineering 3

Because the cycle life of an ultracapacitor can reach morethan 1 million times the effect of the life decay can be ig-nored and the mathematical model is described as follows

Cu nu2Cuc

nu1

Ru nu1Ruc

nu2

Uu nu1Uuc

SOCu Uuc

Uun

Eu 05nu1nu2CucU2unSOC

2u

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(2)

where Cu Ru and Uu respectively represent the capacityinternal resistance and terminal voltage of the ultra-capacitor Cuc Ruc and Uuc respectively represent thecapacity internal resistance and terminal voltage of themonomer nu1 and nu2 respectively represent the number ofseries monomers and parallel modules in the ultracapacitorand Uun is the nominal voltage

3 Control Strategy and Optimization of HESS

31 Fuzzy Control Strategy A top-level model of an EVvehicle is shown in Figure 4 ere are two energy storagedevices in the energy system for EVs the battery andultracapacitorese jointly output power to meet the powerdemand of the vehicle Owing to the complexity of the actualoperation conditions of the vehicle unlike a pure EV with asingle power supply the output power of the battery is usedto cope with all conditions e output power of the batteryand ultracapacitor should be reasonably distributed by thecontrol strategy of HESS to improve the working efficiencyand economy of the vehicle [26]

According to the operation conditions and power de-mand of the EV the demand power of the EV is Preq and theoutput power of the HESS (Phy) is composed of the batteryoutput power (Pbat) ultracapacitor output power (Puc) andsystem loss power (Pe) e mathematical model can bewritten as

Phy Pbat + Puc + Pe

Preq Phy minus Pe(3)

Since the energy loss of the system is low and difficult tocalculate and the loss is ignored in this study the outputpower of the HESS can be calculated as

Preq Phy Pbat + Puc (4)

In the operating process of the vehicle the output powerof the battery pack and ultracapacitor is mainly determinedby the SOC of the batteryultracapacitor and the systemdemand power erefore energy distribution factors (Kbatand Kuc) are proposed to describe the power output of thebattery and capacitor shown as follows

Pbat PreqKbat

Puc PreqKuc

Kbat + Kuc 1

⎧⎪⎪⎨

⎪⎪⎩(5)

Fuzzy control is widely used in various fields For themanagement of vehicle energy system the control methodcan set different control variables and controlled objectsand improve the fuel economy and emissions of the wholevehicle by establishing different fuzzy rules [27ndash29] Basedon the structure and power requirement of the energysystem the structure of fuzzy control logic is shown inFigure 5 A fuzzy logic control strategy is used to managethe energy transport and two fuzzy control rules thatrepresent the output power for driving and recovery powerfor braking are established e fuzzy logic rule for theoutput driving energy adopts the form of three inputs andone output e three inputs are the vehicle demand power(Preq) and the SOC of the battery (SOCbat) and ultra-capacitor (SOCuc) e output is the energy distributionfactor (Kuc) e fuzzy logic rule of braking energy re-covery adopts the form of two inputs and one output etwo inputs are the SOCs of the battery (SOCbat) andultracapacitor (SOCuc) and the output is the energy dis-tribution factor (Kuc)

e construction form can avoid the frequency of highcurrent output from the battery as much as possible on thepremise that the power performance of the vehicle is sat-isfied When there is a high energy demand the ultra-capacitor must have enough energy output power Whencarrying out braking energy recovery the ultracapacitor isused for energy recovery

e Expert Experience Method is used to determine themembership function and the Gauss Z S and Trianglefunctions are selected as the membership functions to es-tablish the fuzzy control rules of the HESS e surface viewof fuzzy rules is shown in Figure 6

I

U+

ndash

Figure 3 Internal-resistance model of the ultracapacitor


32 Algorithm Optimization In order to use the algorithmfor optimization it is necessary to transform the specificproblem into a mathematical model and establish themapping relationship between the value space and coding

that is the coding is used to represent the problem [30]Because there are 27 membership functions in the fuzzycontroller in this paper the Gauss Z and S membershipfunctions need only two variables to determine their

Wheel andaxle ltwhgt

Vehicle ltvehgt

GalTotal fuel used (gal)

Powerbus ltpbgt

Motorcontroller ltmcgt

Gearbox ltgbgtFinal drive ltfdgt

Energystorage ltessgt

Electric accloads ltaccgt

Drive cycleltcycgt

Control system

0No fuel used

[0 0 0 0]No emissions-EV

Version ampCopyright

EmisHC CO NOx (gs)

Ground

TimeGoto ltsdogt

DC-DC

Clockltvcgt evltsdogt ev

UltracapacitorSystem

++

Figure 4 Top-level model of the EV

1Reqrsquod power from

energy storage (W)

0

Constant-K-

GainFuzzy logiccontroller

Fuzzy logiccontroller1

Switch

[SOC]

From

[SOC2]

From1

Product


battery (W)

2Reqrsquod power fromultracapacitor (W)

Add

lt=

Relationaloperator

|u|

Abs

Scope Scope1

Scope2

[SOC]

From2

[SOC2]

From3

times

ndash+

Figure 5 Structure of fuzzy logic control

0806

0402

SOCbat

06

04

02

K uc

0 02 04 06 08 1

Preq

(a)

SOCuc

K uc

08

06

04

021

0806

0402 02

0406

08

SOCbat

(b)

Figure 6 e surface view of fuzzy rules (a) Driving (b) Braking


position and shape while the Triangle function needs threevariables to determine the position and shape of the func-tion erefore 65 parameters are needed to express thevalue space as follows

X x11 x

21 x

116 x

216 x

117 x

217 x

317 x

127 x

227 x

3271113872 1113873

(6)

e algorithm is used for optimization and the math-ematical model of the objective function is described as

miny f(x) (7)

e energy consumption per unit mileage is set as theevaluation standard for the algorithm It is shown as

f(x) fitness energydistance

(8)

e energy consumption of the HESS needs to considerthe consumption of various components including thebattery loss supercapacitor loss DCDC converter loss lineloss and motor loss in which the battery capacitor and DCDC converter are the main consumption objects Otherlosses are ignored e mathematical model is shown asfollows

energy Pbat + Puc + Elbat + El

uc + Eldc

Elbat I2bat(t)R

Eluc I2uc(t)Ruc

Eldc Iin(t) 1 minus ηdc( 1113857

⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(9)

where Pbat and Puc are the output power of the battery packand the ultracapacitor respectively El

bat Eluc and El

dc are theloss of the battery pack ultracapacitor and DCDC con-verter respectively Ibat and Iuc are the working current ofthe battery pack and the ultracapacitor respectively and ηinis the input current of the DCDC converter

To sum up the algorithm optimizes the objectivefunction shown as follows

fitness Pbat + Puc

distance (10)

321 Genetic Algorithm Optimization A genetic algorithm(GA) simulates the evolution phenomenon of the Darwiniantheory of survival of the fittest in nature and uses the processof survival of the fittest and continuous genetic optimizationin the process of evolution to solve the problem and find theoptimal solution All solutions are encoded and the range ofthe solution is constantly close to the optimal solutionthrough generations of genetic operations to solve theproblem Based on the evolutionary characteristics of theGA the inherent properties of the problem are not needed inthe process of searching the solution e ergodicity of theindividual enables the algorithm to effectively carry out aglobal search in the sense of probability and has betteridentification accuracy for the entire world [31 32] eprocess of the GA is shown in Figure 7

First the solution to the specific problem is encoded andthe set of corresponding potential solutions is the initialpopulation Suppose that there are n individuals in an initialpopulation and the corresponding chromosomes and fitnessare shown as

chrom

x11 x1

2

x21 x2

2

xn1 xn

2

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

fitness f1 f2 f3 fn1113858 1113859

(11)

en according to the specific problem differentstrategies are used to evaluate the fitness of individuals andthe offspring is selected according to the fitness Individualswith high fitness are more likely to be selected New pop-ulations are generated through cross recombination andmutation When they are inherited from the selected algebraor meet the fitness requirements the individuals with thehighest fitness output from the current population are takenas the optimal solution

In this paper the control strategy is optimized based onthe GA e steps of building the GA-Fuzzy Control are asfollows

(1) Initialization algorithm set the evolution algebra to80 number of variables to 65 recombinationprobability to 05 mutation probability to 0001 andgeneration gap to 095

Initialpopulation

Calculateindividual

fitness

Proportionaloperation

Crossoveroperation

Mutationoperation

Regenerationpopulation

Finish

Output optimalsolution

Start

End

Y

N

Figure 7 Flowchart of GA


(2) Initialization population 65 individuals are ran-domly generated within the target range as the initialpopulation

(3) Calculate fitness calculate individual fitnessaccording to the fitness function

(4) Judgment condition whether the highest fitness ofthe individual meets the requirements or whetherthe evolutionary algebra is terminated

(5) Update the population select cross over and mutatethe population to generate a new population andreturn to the judgment conditions to continue theevolution process

(6) Save the optimal solution and establish the GA-Fuzzy Control strategy embedded in the EVmodel ofMATLABAdvisor for simulation

322 Particle Swarm Optimization PSO is a type of globalrandom search algorithm based on swarm intelligence Itsimulates the migration and swarm behavior in theprocess of bird swarm foraging When solving specificproblems in the target search space by combining theindividual optimal solution and the group optimal so-lution the optimal solution of the target area is searchediteratively [33ndash36] A flowchart of the PSO is shown inFigure 8

In the D-dimensional target search space the initialpopulation is composed of n particles where the positionand velocity of the ith particle are D-dimensional vectorsshown as follows

Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)

e optimal positions searched by the ith particle and theentire particle swarm are the individual extremum andglobal extremum respectively shown as follows

Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)

After the individual and global extremum are updatedthe particle updates its own speed and position according tothe current position and speed and the distance from theoptimal particle the update rule is

Vid ωvid + c1 random(0 1) pid minus xid( 1113857

+ c2 random(0 1) pgd minus xid1113872 1113873

xid xid + vid

(16)

where ω is the inertia factor (adjusting the global optimi-zation ability and local optimization performance) and c1and c2 are acceleration constants where the former is theindividual learning factor of each particle and the latter is thesocial learning factor of each particle ese are usually set asc1 c2 isin [0 4]

Based on a PSO algorithm to optimize the controlstrategy the steps of building PSO-Fuzzy Control are asfollows

(1) Initialization algorithm set the maximum number ofiterations to 80 number of particles to 65 maximumspeed to 05 and minimum speed to minus05

(2) Initialize particle swarm randomly generate parti-cles with different positions and velocities in thetarget search space

(3) Evaluate particles calculate the fitness of particlesaccording to the evaluation criteria

(4) Update the optimum update the optimal positionexperienced by particles and groups

(5) Judgment condition whether the optimal fitness ofparticles meets the requirements or whether theiterations are terminated

(6) e optimal solution is saved and the PSO-FuzzyControl strategy is embedded into the EV vehiclemodel of MATLABAdvisor for simulation

33DynamicProgramming In order to compare the controlperformance of GA-Fuzzy Control and PSO-Fuzzy Controlmore accurately this paper proposes a dynamic program-ming (DP) algorithm to calculate the theoretical minimumenergy consumption of HESS DP algorithm is usually usedto solve multistage decision-making optimization problemswhich are decomposed into subproblems and solved step bystep Because HESS energy management strategy can beconsidered as a multistage decision-making problem in

Start

Initializeparticle swarm

Evaluateparticles

FinishN

Y


End

Regeneration population

Update individual positionand speed

Update the optimal positionof individuals and group

Figure 8 Flowchart of PSO


discrete time the power output of battery and ultracapacitorcan be regulated in different stages to obtain the best controlperformance erefore DP algorithm is suitable for thebenchmark evaluation method of HESS energy managementstrategy [37]

In this paper the subproblem is to solve the minimumenergy consumption of HESS when the initial state istransferred to the current state variable group In each stagethe solution and optimization are carried out and finally theminimum energy consumption of HESS in each stage isobtained

e optimization objective is shown as follows

Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)

During the optimization process the ultracapacitor SOCis constrained so that the SOC of ultracapacitor at the end isconsistent with that at the initial state e expression is asfollows

Ibatmin le Ibat le Ibatmax

Iucmin le Iuc le Iucmax

02 le SOCuc le 1

SOCucinitial SOCucend

4RbatPuc ref le U2bat

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)

where Puc ref is the theoretical output power of theultracapacitor

e input variable in the optimization process is Pd thestate variable is SOCuc and the decision variable is Puc ref First the theoretical output power of ultracapacitor is cal-culated shown as follows

Euc(i j k m) 05CU2uc SOC2

uc(1 m) minus SOC2uc(1 j)1113872 1113873

Puc ref(i j k m) Euc(i j k m) minus RucEuc(i j k m)

UucSOCuc(1 m)1113888 1113889

2

(19)

According to the system demand power and theoreticaloutput power of ultracapacitor the battery current andenergy consumption are calculated

Ibat(i j k m)

Ubat minus 4RbatPuc ref(i j k m)

1113968

2Rbat

Ebat(i j k m) Ibat(i j k m)Ubat

(20)

At the end of calculation the minimum value can bestore in Econ_ref by comparing the energy consumption ofeach stage

if Econ ref(i + 1 j k)gtEstate(i j k m) + Econ_state

thenEcon ref(i + 1 j k) Estate(i j k m) + Econ_state

⎧⎨

⎩

(21)

After the minimum energy consumption of the system isobtained the optimal allocation mode can also be obtainedthrough path backward pushing

4 Results and Discussion

In order to confirm the effect of algorithm optimization thefuzzy control strategy of a HESS optimized by the GA and PSOalgorithms is examinede improved EVmodel inMATLABAdvisor is used for simulations e following simulationdriving cycles are used UDDS NEDC and ChinaCity Aschematic diagram of the operation is shown in Figure 9 ecycle conditions of the three different countries and regions caneffectively test the performance of the optimized HESS

41 Analysis of Simulation Results To evaluate the batteryprotection performance of the energy management strategy(EMS) optimized by different algorithms as shown in Figure 10the battery working currents for different driving cycles arecompared It can be seen that based on the three conditions theoutput current fluctuation of the battery is more stable in thesimulation process From Table 1 in the conditions of UDDSNEDC andChinaCity the peak current of GA-FuzzyControl islower than that of PSO-Fuzzy Control by 356001 A 199046 Aand 465270 A respectively

In general the economy of the vehicle can be evaluatedby examining the fuel economy of the vehicle As this studyis based on an EV other losses are ignored and the energyconsumption of the HESS is regarded as the economicevaluation standard of the entire vehicle Figure 11 shows thetotal energy consumption of two different strategies forsimulations in various operating conditions It can be seenfrom the figure that in three driving cycles the total energyconsumption when using GA-Fuzzy Control and PSO-Fuzzy Control as energy management strategies is lowerthan that before optimization is verifies the effectivenessof the algorithm optimization

Compared with the data in Table 1 in the operatingconditions of UDDS NEDC and ChinaCity the total energyconsumption of GA-Fuzzy Control decreased by 2448990604 and 25332 respectively compared with that beforeoptimization e energy consumption of PSO-Fuzzy Controldecreased by 10859 09659 and 02650 respectively esimulation results of the two strategies show that the totalenergy consumption of the control strategy optimized by theGA is lower Combined with the comparison results of theworking current of the battery the optimization effect of theGA in terms of protection of the battery and the battery lifestability is better which helps save more energy

Keeping the simulation conditions unchanged thispaper uses the DP algorithm to calculate the theoreticalminimum energy consumption of HESS which is listed inTable 1 for comparison Compared with the theoreticalminimum energy consumption the simulation results ofGA-Fuzzy Control under three drive cycles increased by044 035 052 respectively is proves that thecontrol strategy proposed in this paper is approximately thebest for the optimization of HESS energy consumption

42 Discussion e GA and PSO algorithms have manyfeatures in common After the population is randomlyinitialized both of them use fitness function to evaluate the


UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

Figure 9 Operation diagram of three driving cycles

100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)

GA-Fuzzy ControlPSO-Fuzzy Control

(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)

Figure 10 Battery current for three driving cycles (a) UDDS (b) NEDC and (c) ChinaCity


system and search randomly according to the fitnessfunction

In this paper based on the control strategy of the hybridpower system of the pure electric vehicle under the sameoptimization conditions the fuzzy rules are optimized byusing the GA and PSO By comparing the control effects ofGA-Fuzzy Control and PSO-Fuzzy Control the accuracy of

the two algorithms is evaluatede results show that the GAis more accurate

In [38 39] the two algorithms are also used for researchReference [38] focuses on the optimization of kinetic pa-rameters of biomass pyrolysis e results show that the PSObased on the three-component parallel reaction mechanismof biomass pyrolysis has the advantages of being closer to the

Table 1 Parameters of peak current and energy consumption for different scenarios

Peak current (A) Energy consumption (times106 J)

Fuzzy ControlUDDS 67501NEDC 62845

ChinaCity 57359

GA-Fuzzy ControlUDDS 534040 65848NEDC 629496 57151

ChinaCity 453493 55906

PSO-Fuzzy ControlUDDS 890041 66768NEDC 828542 62238


DPUDDS 65561NEDC 56954

ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106

Fuzzy ControlGA-Fuzzy ControlPSO-Fuzzy Control

0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)

Figure 11 Energy consumption for three driving cycles (a) UDDS (b) NEDC and (c) ChinaCity


global optimal solution and having faster convergence speedthan the GA In [39] two models of algorithms are estab-lished and applied to Iranrsquos oil demand forecast e resultsshow that the demand estimation models of the two algo-rithms are in good agreement with the observed data but thePSO model has the best performance

In the case of binary distribution or discrete distributionof the data in this paper crossover and mutation operationsin the GA are very helpful to find the global optimum andthe effect is better than the gradual approximation of PSO In[40 41] for global optimization with continuous value PSOoptimization has memory function It moves to global andlocal optimal direction in each iteration and can approachthe optimal solution faster It has strong optimizationperformance and fast optimization speed

5 Conclusions

Based on the characteristics of poor life stability and limitedbattery life of an EV as a new energy vehicle this paperstudied an EMS and optimized the management strategy toreduce the energy consumption of the HESS and protect thebattery life

(1) Aimed at the semiactive battery load structure of theHESS of a pure electric vehicle a fuzzy controlstrategy was selected as the power EMS and thecontrol framework was constructed based on an EVmodel in MATLABAdvisor e output power ratioof the battery and ultracapacitor was controlledthrough the vehicle demand power and SOCs of thebattery and ultracapacitor to achieve the purpose ofoptimal management

(2) e energy consumption per unit mileage was set asthe evaluation standard of the algorithm GA andPSO were used to improve the fuzzy control strategyin the software establish the GA-Fuzzy Control andPSO-Fuzzy Control strategies and conduct a sim-ulation based on the operating conditions of UDDSNEDC and ChinaCity e results showed that bothalgorithms can optimize the energy managementand control strategy of the energy system and theGA had better optimization performance e GAshowed better protection and economy in the sim-ulation to meet the requirements for the enduranceof pure electric vehicles equipped with a HESS

(3) e DP algorithm is used as the benchmark methodto calculate the theoretical minimum energy con-sumption of HESS in this simulation environmentCompared with the simulation results of GA-FuzzyControl it is verified that the control strategy pro-posed in this paper is approximately optimal for theoptimization of HESS energy consumption

(4) Both GA and PSO can optimize fuzzy controlstrategies but the time-consuming algorithm is notsuitable for real-time optimizatione optimizationmethod used in this article is only applicable whenthe simulation drive cycle is known in advance

Data Availability

e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

e authors declare no potential conflicts of interest withrespect to the research authorship andor publication ofthis article

Acknowledgments

is research was funded by the National Natural ScienceFoundation of China and Natural Science Foundation ofZhejiang Province (Grant Nos 51741810 7187107871874067 and LGG18E080005)

References

[1] M Neenu and S Muthukumaran ldquoA battery with ultra ca-pacitor hybrid energy storage system in electric vehiclesrdquo inProceedings of the International Conference on Advances inEngineering IEEE Nagapattinam Tamil Nadu India March2012

[2] J Cao and A Emadi ldquoA new batteryUltraCapacitor hybridenergy storage system for electric hybrid and plug-in hybridelectric vehiclesrdquo IEEE Transactions on Power Electronicsvol 27 no 1 pp 122ndash132 2012

[3] S Lu K A Corzine and M Ferdowsi ldquoA new batteryultracapacitor energy storage system design and its motordrive integration for hybrid electric vehiclesrdquo IEEE Trans-actions on Vehicular Technology vol 56 no 4 pp 1516ndash15232007

[4] M E Choi and S W Seo ldquoRobust energy management of abatterysupercapacitor hybrid energy storage system in anelectric vehiclerdquo in Proceedings of the 2012 IEEE InternationalElectric Vehicle Conference IEEE Greenville SC USA March2012

[5] G Wang P Yang and J Zhang ldquoFuzzy optimal control andsimulation of battery-ultracapacitor dual-energy sourcestorage system for pure electric vehiclerdquo in Proceedings of theIntelligent Control and Information Processing (ICICIP) IEEEDalian China August 2010

[6] O Erdinc B Vural and M Uzunoglu ldquoA wavelet-fuzzy logicbased energy management strategy for a fuel cellbatteryultra-capacitor hybrid vehicular power systemrdquo Journal ofPower Sources vol 194 no 1 pp 369ndash380 2009

[7] M Michalczuk B Ufnalski and L Grzesiak ldquoFuzzy logiccontrol of a hybrid battery-ultracapacitor energy storage foran urban electric vehiclerdquo in Proceedings of the 2013 EighthInternational Conference and Exhibition on Ecological Vehiclesand Renewable Energies (EVER) Monte Carlo MonacoMarch 2013

[8] J Y Liang J L Zhang X Zhang et al ldquoEnergy managementstrategy for a parallel hybrid electric vehicle equipped with abatteryultra-capacitor hybrid energy storage systemrdquo Journalof Zhejiang University-Science A (Applied Physics amp Engi-neering) vol 14 no 8 pp 4ndash22 2013

[9] Q Wang S Du L Li et al ldquoReal time strategy of plug-inhybrid electric bus based on particle swarm optimizationrdquoJournal of Mechanical Engineering vol 53 no 4 2017 inChinese


[10] C Yang X Jiao L Li et al ldquoResearch on real-time opti-mization strategy of plug-in hybrid power for bus applica-tionsrdquo Journal of Mechanical Engineering vol 51 no 22pp 117ndash125 2015 in Chinese

[11] M Zandi A Payman J-P Martin S Pierfederici B Davatand F Meibody-Tabar ldquoEnergy management of a fuel cellsupercapacitorbattery power source for electric vehicularapplicationsrdquo IEEE Transactions on Vehicular Technologyvol 60 no 2 pp 433ndash443 2011

[12] M Hajer K Ameni M Jean-Philippe A Mansour P Sergeand B Faouzi ldquoImplementation of energy managementstrategy of hybrid power source for electrical vehiclerdquo EnergyConversion and Management vol 195 pp 830ndash843 2019

[13] L Ji Z Quan H Yinglong et al ldquoDual-loop online intelligentprogramming for driver-oriented predict energymanagementof plug-in hybrid electric vehiclesrdquo Applied Energy vol 253p 113617 2019

[14] LW Chong YWWong R K Rajkumar et al ldquoAn adaptivelearning control strategy for standalone PV system withbattery-supercapacitor hybrid energy storage systemrdquo Journalof Power Sources vol 394 2018

[15] M Montazeri-Gh and A Fotouhi ldquoTraffic condition recog-nition using the -means clustering methodrdquo Scientia Iranicavol 18 no 4 pp 930ndash937 2011

[16] A Fotouhi and M Montazeri-Gh ldquoTehran driving cycledevelopment using the k-means clustering methodrdquo ScientiaIranica vol 20 no 2 pp 286ndash293 2013

[17] M Montazeri A Fotouhi and A Naderpour ldquoDrivingsegment simulation for determination of the most effectivedriving features for HEV intelligent controlrdquo Vehicle SystemDynamics vol 50 no 2 pp 229ndash246 2012

[18] M Montazeri-Gh A Fotouhi and A Naderpour ldquoDrivingpatterns clustering based on driving features analysisrdquo Pro-ceedings of the Institution of Mechanical Engineers Part CJournal of Mechanical Engineering Science vol 225 no 6pp 1301ndash1317 2011

[19] C Zhang Research on the gteory of Composite Power SupplyMatching and Control for Pure Electric Vehicle Jilin Uni-versity Changchun China 2017 in Chinese

[20] A Fotouhi D J Auger K Propp S Longo and M Wild ldquoAreview on electric vehicle battery modelling from Lithium-ion toward Lithium-Sulphurrdquo Renewable and SustainableEnergy Reviews vol 56 pp 1008ndash1021 2016

[21] V H Johnson ldquoBattery performance models in ADVISORrdquoJournal of Power Sources vol 110 no 2 pp 321ndash329 2002

[22] L Qian W Wu and H Zhao ldquoSimulation analysis of batterymodel based on advisor softwarerdquo Computer Simulationno 8 pp 166ndash168 2004 in Chinese

[23] J Cao ldquoBatteryultra-capacitor hybrid energy storage systemfor electric hybrid electric and plug-in hybrid electric vehi-clesrdquo Dissertations amp gteses-Gradworks vol 27 no 1pp 122ndash132 2010

[24] I Urasaki ldquoHybrid power source with electric double layercapacitor and battery in electric vehiclerdquo in Proceedings of theInternational Conference on Electrical Machines amp SystemsIEEE Incheon South Korea October 2010

[25] Y Tang and A Khaligh ldquoOn the feasibility of hybrid BatteryUltracapacitor Energy Storage Systems for next generationshipboard power systemsrdquo in Proceedings of the Vehicle Powerand Propulsion Conference (VPPC) 2010 IEEE Lille FranceSeptember 2010

[26] Z He and Z Fu ldquoFuzzy control strategy for energy man-agement of hybrid electric vehiclerdquo Computer Measurementand Control vol 21 no 12 pp 3256ndash3259 2013 in Chinese

[27] N A Kheir M A Salman and N J Schouten ldquoEmissions andfuel economy trade-off for hybrid vehicles using fuzzy logicrdquoMathematics and Computers in Simulation vol 66 no 2-3pp 155ndash172 2004

[28] N J Schouten M A Salman and N A Kheir ldquoFuzzy logiccontrol for parallel hybrid vehiclesrdquo IEEE Transactions onControl Systems Technology vol 10 no 3 pp 460ndash468 2002

[29] N J Schouten M A Salman and N A Kheir ldquoEnergymanagement strategies for parallel hybrid vehicles using fuzzylogicrdquo Control Engineering Practice vol 11 no 2 pp 171ndash1772003

[30] X Song and X Ren ldquoEnergy management strategy of FSMfuzzy hybrid electric vehicle based on improved artificial beecolony algorithmrdquo Modern Manufacturing Engineeringno 03 pp 68ndash119 2018 in Chinese

[31] S G Li S M Sharkh F C Walsh and C N Zhang ldquoEnergyand battery management of a plug-in series hybrid electricvehicle using fuzzy logicrdquo IEEE Transactions on VehicularTechnology vol 60 no 8 pp 3571ndash3585 2011

[32] G Narges K Alibakhsh T Ashkan B Leyli and M AminldquoOptimizing a hybrid wind-PV-battery system using GA-PSOand MOPSO for reducing cost and increasing reliabilityrdquoEnergy vol 154 pp 581ndash591 2018

[33] J J Liang and P N Suganthan ldquoDynamic multi-swarmparticle swarm optimizer with local searchrdquo in Proceedings ofthe 2005 IEEE Congress on Evolutionary Computation IEEEHong Kong China June 2005

[34] C Syuan-Yi W Chien-Hsun H Yi-Hsuan and C Cheng-TaldquoOptimal strategies of energy management integrated withtransmission control for a hybrid electric vehicle using dy-namic particle swarm optimizationrdquo Energy vol 160pp 154ndash170 2018

[35] N Bounar S Labdai and A Boulkroune ldquoPSO-GSA basedfuzzy sliding mode controller for DFIG-based wind turbinerdquoISA Transactions vol 85 pp 177ndash188 2019

[36] H Zhang and H Qing ldquoParallel multiagent coordinationoptimization algorithm implementation evaluation andapplicationsrdquo IEEE Transactions on Automation Science andEngineering vol 14 no 2 pp 984ndash995 2016

[37] C Song F Zhou and F Xiao ldquoEnergy management opti-mization of composite power supply based on dynamicplanningrdquo Journal of Jilin University Engineering Editionvol 47 no 188 pp 8ndash14 2001 in Chinese

[38] Y Ding W Zhang L Yu and K Lu ldquoe accuracy andefficiency of GA and PSO optimization schemes on estimatingreaction kinetic parameters of biomass pyrolysisrdquo Energyvol 176 no 1 pp 582ndash588 2019

[39] E Assareh M A Behrang M R Assari andA Ghanbarzadeh ldquoApplication of PSO (particle swarm op-timization) and GA (genetic algorithm) techniques on de-mand estimation of oil in Iranrdquo Energy vol 35 no 12pp 5223ndash5229 2010

[40] H Zhang ldquoA discrete-time switched linear model of theparticle swarm optimization algorithmrdquo Swarm and Evolu-tionary Computation vol 52 p 100606 2020

[41] N Kassarwani J Ohri and A Singh ldquoPerformance analysis ofdynamic voltage restorer using improved PSO techniquerdquoInternational Journal of Electronics vol 106 no 2 pp 212ndash236 2019


connections which solves the problem of large voltagechanges caused by ultracapacitor power transmission usthe drive performance of the motor is not disturbed

Research on energy management strategies can optimizethe energy efficiency of the entire vehicle without changingthe basic components and framework of the HESS In [4] anoptimization framework was proposed to calculate thesuboptimal current of the hybrid system in an EV to controlthe current and minimize the working current and fluctu-ation of the battery in the EV In [5ndash8] by combining abattery and ultracapacitor an energy control strategy withfuzzy control strategy as the core was proposed to improvethe fuel economy and durability of energy system compo-nents while maintaining the vehicle power performance In[9 10] a real-time optimal energy management strategy(EMS) was proposed for a plug-in hybrid bus based on theminimum equivalent fuel consumption strategy and con-sidering the frequent starting of a low-speed engine In[11 12] the energy management of a hybrid power systemwas analyzed e energy flow of different energy sourceswas managed separately by combining fuzzy logic controland shape control In [13] the online predictive controlstrategy of a series and parallel plug-in hybrid EV wasstudied A new dual-loop online intelligent planning (DOIP)method for speed prediction and energy flow control wasproposed and a depth fuzzy predictor was established torealize directional speed prediction e optimal controlbehavior was determined by learning the vehicle speed andacceleration

ere are also some scholars who use algorithms tooptimize research Reference [14] has proposed an adaptivepower distribution algorithm e parameters of the algo-rithm were optimized by combining self-organizing map-ping and particle swarm optimization (PSO) to alleviate thepeak demand and short charge and discharge period of thebattery In [15 16] the focus is on the k-means clusteringalgorithm Reference [15] focuses on the study of the drivingcycle of cars in Tehran and its suburbs By collecting thesituation of vehicles running in actual traffic the calculationis based on the definition of ldquomicro-travelrdquo Two drivingfunctions ldquoaverage speedrdquo and ldquoidle time percentagerdquo arecalculated e micro-trips are then clustered into fourgroups in driving feature space using the k-means clusteringmethod e focus of [16] is the application of drivingcondition recognition in hybrid electric vehicle intelligentcontrol For this purpose driving features are identified andused for driving segment clustering using the k-meansclustering algorithm Many combinations of driving featuresand different numbers of clusters are evaluated in order toachieve the best traffic condition recognition results Ref-erences [17 18] also record the vehicle data and driving dataunder the actual traffic conditions analyze the driving endand study the influence of driving characteristics on the fuelconsumption and exhaust emissions through the drivingsegment simulation e results of [17] show that the ve-locity-dependent driving features such as ldquoenergyrdquo ldquomean ofvelocityrdquo ldquodisplacementrdquo and ldquomaximum velocityrdquo aremore effective in vehicle exhaust emissions and fuel econ-omy e results of [18] show that ldquoenergyrdquo and ldquopercentage

of free timerdquo are two driving characteristics to drive seg-mented clustering Driving segment clustering can be usedin driving cycle development intelligent hybrid vehiclecontrol and so on

In the current study most of the control strategies of EVsare optimized in one algorithm Based on the above researchthis paper uses fuzzy control strategy as an EMS with thecomposite power supply form of a combined battery andultracapacitor Reducing the total energy consumption is theoptimization goal and GA and PSO are used to carry outlearning optimization for the fuzzy control strategy inMATLABAdvisor in the UDDS NEDC and ChinaCitydriving cycles In the same simulation environment thetheoretical minimum energy consumption of HESS is cal-culated by DP algorithm An experimental analysis is carriedout to compare the battery current output performance andthe total energy consumption parameters of the energysystem and to evaluate the optimization effect of thealgorithm

2 System Description and Methodology

21 Modeling of HESS ere are three common types ofcomposite power supply structure passive semiactive andactive e semiactive structure has two types battery endload and capacitor end load e main feature of the passivehybrid energy system is that the battery and the ultra-capacitor are directly connected in parallel on the power busand the voltage of the two is synchronized in real time suchthat the energy distribution between the battery andultracapacitor cannot be flexibly adjusted in the case of high-power output or full capacity e composite power supplywith an active structure has two DCDC converters that areconnected in series with batteries and capacitors e twoDCDC converters are integrated and connected in seriesBoth of these forms can flexibly adjust the distribution of theenergy output but the energy system structure is complexand the actual application cost is high [19]

erefore in this study the battery terminal loadstructure of a semiactive energy system is selected Its to-pology is shown in Figure 1 e battery and capacitor areisolated by a DCDC converter e battery acts as the mainpower supply and the capacitor is the auxiliary powersupply e energy system distribution of the battery andcapacitor can be realized through high and low voltageconversions of the voltage Compared with the battery endload the ultracapacitor is used as the main power supply inthe form of the capacitor end load Owing to the low capacityof the capacitor energy the battery needs to participate indriving and all current flowing through the battery needs topass through the DCDC converter is makes the batterywork with low efficiency and high energy consumption

22 Modeling of Battery Battery modeling is a significanttask within battery technology development and is vital inapplications For example EV range prediction is onlypossible through the application of advanced batterymodeling and estimation techniques to determine current






Qb nb1nb2Qbc

Rb nb1Rbc

nb2

Ub nb1Ubc

dSOCb

dt

Ub minusU2

b minus 4RbPm

1113969

2RbQb

Pb minusdSOCb

dtUbQb

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(1)





Battery

Motor controller

Driving motorLoad


I

U

+

ndash




Cu nu2Cuc

nu1

Ru nu1Ruc

nu2

Uu nu1Uuc

SOCu Uuc

Uun


2u

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(2)





Phy Pbat + Puc + Pe





Pbat PreqKbat

Puc PreqKuc

Kbat + Kuc 1

⎧⎪⎪⎨

⎪⎪⎩(5)




I

U+

ndash






Vehicle ltvehgt


Powerbus ltpbgt





Drive cycleltcycgt

Control system

0No fuel used



EmisHC CO NOx (gs)

Ground

TimeGoto ltsdogt

DC-DC



++



energy storage (W)

0

Constant-K-



Switch

[SOC]

From

[SOC2]

From1

Product


battery (W)


Add

lt=

Relationaloperator

|u|

Abs

Scope Scope1

Scope2

[SOC]

From2

[SOC2]

From3

times

ndash+


0806

0402

SOCbat

06

04

02

K uc

0 02 04 06 08 1

Preq

(a)

SOCuc

K uc

08

06

04

021

0806

0402 02

0406

08

SOCbat

(b)




X x11 x

21 x

116 x

216 x

117 x

217 x

317 x

127 x

227 x

3271113872 1113873

(6)


miny f(x) (7)



(8)



uc + Eldc

Elbat I2bat(t)R

Eluc I2uc(t)Ruc


⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(9)


bat Eluc and El



fitness Pbat + Puc

distance (10)



chrom

x11 x1

2

x21 x2

2

xn1 xn

2

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

fitness f1 f2 f3 fn1113858 1113859

(11)




Initialpopulation

Calculateindividual

fitness


Crossoveroperation

Mutationoperation


Finish


Start

End

Y

N










Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)


Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)




xid xid + vid

(16)










Start


Evaluateparticles

FinishN

Y


End









Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References
















































Qb nb1nb2Qbc

Rb nb1Rbc

nb2

Ub nb1Ubc

dSOCb

dt

Ub minusU2

b minus 4RbPm

1113969

2RbQb

Pb minusdSOCb

dtUbQb

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(1)





Battery

Motor controller

Driving motorLoad


I

U

+

ndash




Cu nu2Cuc

nu1

Ru nu1Ruc

nu2

Uu nu1Uuc

SOCu Uuc

Uun


2u

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(2)





Phy Pbat + Puc + Pe





Pbat PreqKbat

Puc PreqKuc

Kbat + Kuc 1

⎧⎪⎪⎨

⎪⎪⎩(5)




I

U+

ndash






Vehicle ltvehgt


Powerbus ltpbgt





Drive cycleltcycgt

Control system

0No fuel used



EmisHC CO NOx (gs)

Ground

TimeGoto ltsdogt

DC-DC



++



energy storage (W)

0

Constant-K-



Switch

[SOC]

From

[SOC2]

From1

Product


battery (W)


Add

lt=

Relationaloperator

|u|

Abs

Scope Scope1

Scope2

[SOC]

From2

[SOC2]

From3

times

ndash+


0806

0402

SOCbat

06

04

02

K uc

0 02 04 06 08 1

Preq

(a)

SOCuc

K uc

08

06

04

021

0806

0402 02

0406

08

SOCbat

(b)




X x11 x

21 x

116 x

216 x

117 x

217 x

317 x

127 x

227 x

3271113872 1113873

(6)


miny f(x) (7)



(8)



uc + Eldc

Elbat I2bat(t)R

Eluc I2uc(t)Ruc


⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(9)


bat Eluc and El



fitness Pbat + Puc

distance (10)



chrom

x11 x1

2

x21 x2

2

xn1 xn

2

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

fitness f1 f2 f3 fn1113858 1113859

(11)




Initialpopulation

Calculateindividual

fitness


Crossoveroperation

Mutationoperation


Finish


Start

End

Y

N










Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)


Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)




xid xid + vid

(16)










Start


Evaluateparticles

FinishN

Y


End









Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References













































Cu nu2Cuc

nu1

Ru nu1Ruc

nu2

Uu nu1Uuc

SOCu Uuc

Uun


2u

⎧⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎪⎩

(2)





Phy Pbat + Puc + Pe





Pbat PreqKbat

Puc PreqKuc

Kbat + Kuc 1

⎧⎪⎪⎨

⎪⎪⎩(5)




I

U+

ndash






Vehicle ltvehgt


Powerbus ltpbgt





Drive cycleltcycgt

Control system

0No fuel used



EmisHC CO NOx (gs)

Ground

TimeGoto ltsdogt

DC-DC



++



energy storage (W)

0

Constant-K-



Switch

[SOC]

From

[SOC2]

From1

Product


battery (W)


Add

lt=

Relationaloperator

|u|

Abs

Scope Scope1

Scope2

[SOC]

From2

[SOC2]

From3

times

ndash+


0806

0402

SOCbat

06

04

02

K uc

0 02 04 06 08 1

Preq

(a)

SOCuc

K uc

08

06

04

021

0806

0402 02

0406

08

SOCbat

(b)




X x11 x

21 x

116 x

216 x

117 x

217 x

317 x

127 x

227 x

3271113872 1113873

(6)


miny f(x) (7)



(8)



uc + Eldc

Elbat I2bat(t)R

Eluc I2uc(t)Ruc


⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(9)


bat Eluc and El



fitness Pbat + Puc

distance (10)



chrom

x11 x1

2

x21 x2

2

xn1 xn

2

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

fitness f1 f2 f3 fn1113858 1113859

(11)




Initialpopulation

Calculateindividual

fitness


Crossoveroperation

Mutationoperation


Finish


Start

End

Y

N










Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)


Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)




xid xid + vid

(16)










Start


Evaluateparticles

FinishN

Y


End









Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References















































Vehicle ltvehgt


Powerbus ltpbgt





Drive cycleltcycgt

Control system

0No fuel used



EmisHC CO NOx (gs)

Ground

TimeGoto ltsdogt

DC-DC



++



energy storage (W)

0

Constant-K-



Switch

[SOC]

From

[SOC2]

From1

Product


battery (W)


Add

lt=

Relationaloperator

|u|

Abs

Scope Scope1

Scope2

[SOC]

From2

[SOC2]

From3

times

ndash+


0806

0402

SOCbat

06

04

02

K uc

0 02 04 06 08 1

Preq

(a)

SOCuc

K uc

08

06

04

021

0806

0402 02

0406

08

SOCbat

(b)




X x11 x

21 x

116 x

216 x

117 x

217 x

317 x

127 x

227 x

3271113872 1113873

(6)


miny f(x) (7)



(8)



uc + Eldc

Elbat I2bat(t)R

Eluc I2uc(t)Ruc


⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(9)


bat Eluc and El



fitness Pbat + Puc

distance (10)



chrom

x11 x1

2

x21 x2

2

xn1 xn

2

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

fitness f1 f2 f3 fn1113858 1113859

(11)




Initialpopulation

Calculateindividual

fitness


Crossoveroperation

Mutationoperation


Finish


Start

End

Y

N










Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)


Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)




xid xid + vid

(16)










Start


Evaluateparticles

FinishN

Y


End









Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References













































X x11 x

21 x

116 x

216 x

117 x

217 x

317 x

127 x

227 x

3271113872 1113873

(6)


miny f(x) (7)



(8)



uc + Eldc

Elbat I2bat(t)R

Eluc I2uc(t)Ruc


⎧⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎩

(9)


bat Eluc and El



fitness Pbat + Puc

distance (10)



chrom

x11 x1

2

x21 x2

2

xn1 xn

2

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣

⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦

fitness f1 f2 f3 fn1113858 1113859

(11)




Initialpopulation

Calculateindividual

fitness


Crossoveroperation

Mutationoperation


Finish


Start

End

Y

N










Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)


Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)




xid xid + vid

(16)










Start


Evaluateparticles

FinishN

Y


End









Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References



















































Xi xi1 xi2 xi3 xiD( 1113857 i 1 2 3 n (12)

Vi vi1 vi2 vi3 viD( 1113857 i 1 2 3 i (13)


Pb pi1 pi2 pi3 piD( 1113857 i 1 2 3 n (14)

Gb pg1 pg2 pg3 pgD1113872 1113873 (15)




xid xid + vid

(16)










Start


Evaluateparticles

FinishN

Y


End









Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References















































Econ min1113944t1

Eb(t) + Eu(t)( 1113857 (17)




02 le SOCuc le 1



⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎪⎪⎪⎪⎩

(18)




uc(1 m) minus SOC2uc(1 j)1113872 1113873


UucSOCuc(1 m)1113888 1113889

2

(19)


Ibat(i j k m)


1113968

2Rbat


(20)




⎧⎨

⎩

(21)










UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References












































UDDS

NEDC

ChinaCity

30

25

20

15

10

5

0

Spee

d (k

mh

)

40

30

20

10

0

Spee

d (k

mh

)Sp

eed

(km

h)

20

15

10

5

0

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)

0 200 400 600 800 1000 1200 1400Time (s)


100

80

60

40

20

0

ndash20

ndash400 200 400 600 800 1000 1200 1400

Time (s)

Batte

ry cu

rren

t (A

)


(a)

100

80

60

40

20

0

ndash20

Batte

ry cu

rren

t (A

)

0 200 400 600 800 1000 1200Time (s)


(b)100

50

ndash50

0

0 200 400 600 800 1000 1200 1400Time (s)

Batte

ry cu

rren

t (A

)


(c)










ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References



















































ChinaCity 57359






ChinaCity 55619

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200 1400Time (s)

(a)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)

times106


0 200 400 600 800 1000 1200Time (s)

(b)

6

4

2

0

Ener

gy co

nsum

ptio

n (J

)


0 200 400 600 800 1000 1200 1400Time (s)

times106

(c)





5 Conclusions






Data Availability




Acknowledgments


References














































5 Conclusions






Data Availability




Acknowledgments


References












































OptimizationofHybridEnergyStorageSystemControl...

Documents

Transcript of OptimizationofHybridEnergyStorageSystemControl...