Journal of Power Sources 194(2009)369–380
Contents lists available at ScienceDirect
Journal of Power
Sources
j o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /j p o w s o u
r
A wavelet-fuzzy logic bad energy management strategy for a fuel cell/battery/ultra-capacitor hybrid vehicular power system
O.Erdinc,B.Vural,M.Uzunoglu ∗
Department of Electrical Engineering Yildiz Technical University,Istanbul 34349,Turkey
a r t i c l e i n f o Article history:
Received 16March 2009
Received in revid form 26April 2009Accepted 27April 2009
Available online 3May 2009Keywords:Battery Fuel cell Fuzzy logic
Hybrid electric vehicle Ultra-capacitor Wavelet transform
a b s t r a c t
Due to increasing concerns on environmental pollution and depleting fossil fuels,fuel cell (FC)vehicle technology has received considerable attention as an alternative to the conventional vehicular systems.However,a FC system combined with an energy storage system (ESS)can display a preferable performance for vehicle propulsion.As the additional ESS can fulfill the transient power demand fluctuations,the fuel cell can be downsized to fit the average power demand without facing peak loads.Besides,braking energy can be recovered by the ESS.This study focus on a vehicular system powered by a fuel cell and equipped with two condary energy storage devices:battery and ultra-capacitor (UC).However,an advanced energy management strategy is quite necessary to split the power demand of a vehicle in a suitable way for the on-board power sources in order to maximize the performance while promoting the fuel economy and endurance of hybrid system components.In this study,a wavelet and fuzzy logic bad energy management strategy is propod for the developed hybrid vehicular system.Wavelet transform has great capability for analyzing signals consisting of instantaneous changes like a hybrid electric vehicle (HEV)power demand.Besides,fuzzy logic has a quite suitable structure for the control of hybrid systems.The mathematical and electrical models of the hybrid vehicular system are developed in detail and simulated using MATLAB ®,Simulink ®and SimPowerSystems ®environments.
©2009Elvier B.V.All rights rerved.
1.Introduction
The considerable increa in global warming and the desire to decrea the dependence on depleting fossil fuels in order to ensure an uninterrupted energy supply have led to much rearch in alter-native energy sources,recently.Specifically,transportation area has a major position in energy consumption and the greenhou gas emissions causing global warming.Therefore,interest in new solutions for the replacement of conventional internal combustion engine (ICE)bad propulsion systems in vehicular applications has incread steadily.Among alternative powertrains,fuel cell (FC)technologies have been propod as a potential and attractive solution for vehicular applications due to providing environment friendly operation with the usage of renewable fuel [1].Particularly,the Proton Exchange Membrane FC (PEMFC)emerges as one of the most promising candidates for electric vehicle systems thanks to its simplicity,viability,quick start up,higher power density,relatively high electrical efficiency compared to an approximately sized ICE,and operation at lower temperatures [2].
∗Corresponding author.Tel.:+902123832447;fax:+902122594869.E-mail address:uzunoglu@ ,mehmet (M.Uzunoglu).
However,the unsteady operation of a vehicle may not be appro-priate for the usage of a sole FC system.The power demand of a vehicle motor undergoes significant variations due to acceleration,changes in road surface and traffic conditions.Thus,a sole FC sys-tem may not totally satisfy the power demand of a vehicle by the reason of the limitation of fuel cells to track fast load variations due to their slow respon dynamics.Besides,load demand fluc-tuations in vehicle operation may cau fuel starvation,flooding,membrane drying,and pressure imbalance across the FC mem-brane,which will damage the FC stack and decreas its lifetime [3].Moreover,if the FC system alone supplies all power demand,it would increa the size and cost of the FC system as well as hydrogen consumption.Lastly,commercially available fuel cells are not reversible and they do not have capability to recycle the braking energy.Therefore,hybridizing FC system with an energy storage system (ESS)decreas system cost,improves dynamic performance of overall vehicle system,promotes FC lifetime and provides fuel economy owing to regenerative braking energy capturing.
In many hybrid applications,batteries are utilized as ESS [4].Among the various existing rechargeable batteries,lithium-ion batteries appear to occupy a prime position in various aspects [5].However,in spite of providing a significantly high energy density potential,commercially available battery systems prent some drawbacks,such as low cycle-life,long recharging time and
0378-7753/$–e front matter ©2009Elvier B.V.All rights rerved.doi:10.1016/j.jpowsour.2009.04.072
370O.Erdinc et al./Journal of Power Sources194(2009)369–380
low power densities.Besides,rapidly changing current values may damage the chemical structure of batteries.Thus,the usage of a battery system alone as ESS in an automotive powertrain may not be efficient adequately.As an alternative to the battery systems,ultra-capacitors(UCs)–also known as super-capacitors or double-layer capacitors–have drawn attention for u as ESS [6].UCs prent considerably higher power densities than that of batteries,and extremely higher energy densities than that of conventional electrolytic capacitors.The capability for delivering high power/current values in a significantly short time without facing a structural damage is a key advantage over available battery technologies.However,the energy densities of UCs are significantly lower than battery systems[7].Thus,a battery/ultra-capacitor hybrid ESS for vehicular applications can combine the merits of high energy density of lithium-ion batteries with the high power density of ultra-capacitors to completely meet the vehicle requirements[8].
Owing to the fact that FC,battery and UC have different fea-tures and dynamic characteristics,an ove
rall energy management strategy should be designed for the system to coordinate the power flows among the different energy sources.First of all,the power demand of a HEV should be shared between the available on-board power sources considering their individual dynamics.To develop such a load sharing algorithm for FC,battery and UC hybrid struc-ture,we employed wavelet transform owing to its great capability for analyzing and capturing the transients in an applied signal like HEV power demand[9,10].Thus,the negative effects of instanta-neous changes on the electrochemical structures of FC and battery can be prevented.Despite the fact that the wavelet transform may perform an effective performance as a load sharing algorithm, providing an energy management system bad only on wavelet transform may not be sufficient enough for regulating all the sys-tem powerflow.Considering the SOC values of battery and UC,FC system output power should be decread when both sources have enough charge.Besides,FC system should supply more power if both ESSs have a low SOC level.Thus,the overall system powerflow should be controlled in order to increa fuel economy and promote system performance.Thus,a control strategy must be implemented for the developed energy management strategy,so as to guarantee the hybrid system prosperity.For the control of the hybrid sys-tem,fuzzy logic controller(FLC)is utilized considering the fact that FLC has a quite suitable structure for the control of hybrid energy systems as ud in many studies[4,8,11].Besides,in order to model the power demand of a HEV and
test the performance of the developed energy management algorithm,urban dynamome-ter driving schedule(UDDS)data is ud.Using this dynamic load profile,the hybrid system performance can be evaluated effectively.
In this paper,an energy management system compod of a wavelet-fuzzy logic bad load sharing and control algorithm is implemented for the dynamic PEMFC/Battery/UC hybrid vehicular power system model.This paper is organized as follows.Section2 describes the modeling of PEMFC,lithium-ion battery and UC,and clarifies the wavelet bad load sharing and fuzzy logic bad con-trol algorithm developed for the hybrid system.Section3prents the simulation results of the hybrid vehicular system.Then,conclu-sions are given in Section4.
2.System description and methodology
2.1.Modeling of a PEMFC
PEMFC in a hybrid vehicular system is the primary power source and should be operated for supplying the steady state load demand. The FC system model parameters utilized in the model develop-ment are as follows:A Activation area[cm2(cell−1)]
B Constant utilized in modeling of concentration overvoltage
(V)
C FC double-layer equivalent capacitance(F)
C O
2
Dissolved oxygen concentration in the interface of the
cathode catalyst
E Nernst Nernst instantaneous voltage(V)
F Faraday constant[C(kmol)−1]
I FC FC current(A)
J Current density(A(cm)−2)
J max Maximum current density[A(cm)−2]
N s Number of ries FCs in the stack
N p Number of FC stacks
P H
2
Hydrogen partial pressure(atm)
P O
2
Oxygen partial pressure(atm)《春》原文
q H
2
Amount of hydrogenflow required to meet the load change
[kmol(s)−1]
r m The resistivity of Nafion ries proton exchanging
membrane[ (cm)2m−1]
R a Equivalent resistance reprenting the sum of the
activation and concentration resistances( )
R c The contact resistance between the membrane and
electrodes( )
R m Equivalent membrane impedance( )
R ohmic FC internal ohmic resistance( )
T FC temperature(◦K)
T0,T rt,T ic,T it Empirical parameters utilized for modeling the variation of
FC temperature
U Utilization factor
V act Activation overvoltage(V)
V d Potential drop on R a(V)
V conc Concentration overvoltage(V)
V ohmic Ohmic overvoltage(V)
V FC FC voltage(V)
V Stack Stack voltage(V)
The water content of the membrane
1, 2, 3, 4Constants utilized in modeling of activation overvoltage
The ideal standard potential of a hydrogen/oxygen FC under standard state conditions(25◦C and1at.)
is1.229V with liquid water product.However,the actual FC potential is lower than the ideal potential value due to the irreversible voltage drops occurring in FC systems.There are three types of irreversible volt-age drops,namely activation overvoltage,ohmic overvoltage and concentration overvoltage.At low current densities,the activa-tion overvoltage is responsible for the voltage drop of the FC. Besides,concentration overvoltage becomes more significant at high current densities.The effects of the voltage drops and the corresponding variance of FC voltage can be en from the FC polar-ization curve in Fig.1.The most efficient operating region for FC is the linear region corresponding to cell voltage between0.55V
and
Fig.1.Typical FC polarization curve.
O.Erdinc et al./Journal of Power Sources 194(2009)369–380371
0.8V approximately,as en from Fig.1[2,11].Operating FC in this linear region is quite important for the overall system efficiency.
Considering the effects of the above mentioned irreversible volt-age drops,the output voltage of FC can fundamentally be express as [12,13]
V FC =E Nernst −V act −V conc −V ohmic
(1)
The Nernst’s instantaneous voltage can be written as [14,15]E Nernst =1.229−(8.5×10−4)(T −298.15)
+4.308×10−5×T ×ln
P H 2+
12P O 2
.
(2)
Here,partial pressures of hydrogen and oxygen are considered to be changeable between predefined upper and lower limits for anode and cathode pressures inverly proportional to the molar flows of hydrogen and oxygen,and accordingly FC current.The molar flow of hydrogen that reacts in order to meet the load change can be found as q H 2=
I FC ×N S ×N P
2FU
(3)
Besides,the variation of the FC temperature expression in Eq.(2)can be calculated as [16]
T =273+T 0+(T 0−T rt +T ic ×I FC )
1−exp −
t ×I FC
T it
(4)
The following expression gives activation overvoltage occurring in
FC system as [17,18]
V act = 1+ 2T + 3T (ln(I FC ))+ 4(ln(C O 2))
(5)
In Eq.(5),the concentration of dissolved oxygen at the gas/liquid interface can be defined by a Henry’s law expression of the form [17,19]C O 2=
P O 2
5.08×106exp(−498/T )
(6)
Ohmic overvoltage in FC systems is the measure of the I ×R voltage drop associated with the proton conductivity of the solid polymer electrolyte and electronic internal resistances.Thus,ohmic over-voltage in FC can be reprented as [12]V ohmic =I FC ×R ohmic =I FC ×(R m +R c ).
(7)
The equivalent membrane impedance can be expresd in Ohm’s law as R m =
r m ×
A
.(8)
The resistivity of Nafion ries proton exchange membrane in Eq.(8)can be calculated as [12,14]r m
=
181.6[1+0.03×J +0.062×(T/303)2
×J 2.5]
兰花种植方法[ −0.634−3×J ]exp[4.18×(T −303/T )]
.(9)
where J =
I FC
A
.(10)
Concentration overvoltage in FC systems is caud by the mass transportation which in turn affects the concentration of the hydrogen and oxygen at high current densities.The concentration overvoltage of FC can be expresd as [13]V conc =B ×ln
1−
J J max
.(11)
One important electrochemical phenomenon linking the cell voltage to load current variations in FC systems is the charge
double
Fig.2.FC electrical equivalent circuit [17].
layer capacitor effect.The FC equivalent electrical circuit consider-ing the charge double layer capacitor is shown in Fig.2[17].This electrical capacitor reprents the layer of charge on or near the electrode-electrolyte interface,which is a store of electrical charge and energy [18].Considering the effects of this double layer capac-itor effect,the dynamics of the FC output voltage can be obtained more accurately.
The variation of the resistance R a in Fig.2,reprenting the sum of the activation resistance and concentration resistance,can be calculated as [18]R a =
V act +V conc
I FC
(12)
离石一中
According to the aforementioned double layer capacitor effect,the differential equation related with the voltage drop on R a can be written as [17,19]dV d dt
=I FC
C −V d R a C (13)
Considering the combined effects of thermodynamics,mass transport,kinetics,and ohmic resistance,the variation of the FC output voltage can be calculated as follows [18,19]V FC =E Nernst −V d −V ohmic
(14)
Finally,the voltage of an FC stack formed by N s cells connected in ries can be found as V Stack =N s V FC .
(15)
Fig.3shows the model of the PEMFC bad on Eqs.(1)–(15),which is then embedded into the SimPowerSystems of MATLAB as a controlled voltage source and then integrated into the overall system.
2.2.Modeling of battery pack
In this study,a lithium-ion battery pack is utilized both for sup-plying a portion of the ba load together with FC and capturing the braking energy together with UC.In this ction,the dynamic model of the lithium-ion battery is introduced.The lithium-ion battery model parameters ud in propo
d model are as follows:
C bat Battery capacity (Ah)I bat
ca工具Battery current (A)SOC bat
强的成语
Battery state of charge
循环小数教学反思SOC bat init Initial battery state of charge V bat Battery output voltage (V)
V OC Battery open-circuit voltage (V)
Z eq
Battery equivalent internal impedance ( )
The battery output voltage can be calculated due to the battery open circuit voltage and voltage drop resulting from the battery equivalent internal impedance.Accordingly,the battery output voltage may be expresd as V bat =V OC −i bat Z eq
(16)
The battery open circuit voltage is the difference of the electrical potential between the two terminals of a battery,when there is no external load connected.As the value of battery open circuit voltage
372O.Erdinc et al./Journal of Power Sources 194(2009)
海尔芙拉
369–380
Fig.3.Dynamic model of the FC system.
is strongly dependent on battery SOC,it can be calculated as [20]V OC (SOC bat )=−1.031×exp(−35×SOC bat )+3.685
+0.2156×SOC bat −0.1178×SOC 2
bat +0.321×SOC 3
bat
(17)
The battery SOC can be expresd as
SOC bat =SOC bat
init
−
i bat /C bat dt.(18)
The battery equivalent internal impedance in Eq.(16)consists of a ries resistor (R ries ),and two RC networks compod of R Transient S ,C Transient S ,R Transient L and C Transient L ,as shown in Fig.4.R ries is responsible for the instantaneous voltage drop in battery terminal voltage.The components of RC networks are responsible for short and long-time transients in battery internal impedance.The values of R ries ,R Transient S ,C Transient S ,R Transient L and C Transient L depending on battery SOC can be calculated due to the experimen-tally derived empirical equations given in Ref.[20].
The lithium-ion battery model ud in the hybrid system is obtained by implementing the above mentioned equations in MAT-LAB&Simulink environment.2.3.Modeling of UC bank
The natural structure of UC is appropriate to meet transient and instantaneous peak power demands.The UC bank is ud to pro-vide the difference between the load demand and the FC
system
Fig.4.Battery electrical equivalent circuit [20]
.Fig.5.UC electrical equivalent circuit.
output power.Fig.5shows the electrical equivalent circuit ud in the UC model development.The developed model is verified by experimental studies using a Maxwell ®430F,16V UC,which is cur-rently operational in the authors’lab.Detailed description about the modeling of the UC system can be found in [21],which is a previous study of the authors.2.4.Drive cycle
In order to evaluate the dynamic respon of a developed methodology and make suggestions through the development
in
Fig.6.Power demand of a vehicle according to UDDS cycle.
2021感动中国人物O.Erdinc et al./Journal of Power Sources 194(2009)369–380
373
Fig.7.A schematic diagram of the developed energy management strategy.
Table 1
Specific characteristics of UDDS cycle.Time 1369(s)
Distance 7.45(miles)Max.speed 56.7(mph)Avg.speed 19.6(mph)
Max.accel. 1.48[m (s 2)−1]Max.decel.−1.48[m (s 2)−1]Avg.accel.0.51[m (s 2)−1]Avg.decel.0.58[m (s 2)−1]Idle time
259(s)
the overall structure,standard drive cycles can be utilized.To model the power demand of a vehicular system,UDDS cycle is ud.UDDS cycle is a 1369-s test procedure designed to reflect typical speeds and accelerations in city traffic conditions in the United States.Fig.6shows the power demand in UDDS cycle as a func-tion of time.Specific characteristics of the UDDS cycle are shown in Table 1.
2.5.Wavelet and fuzzy logic bad energy management system In this subction,wavelet and fuzzy logic bad energy man-agement strategy developed for the hybrid vehicular system is introduced.A schematic diagram about the developed energy man-agement system is shown in Fig.7.
For better understanding of the power management strategies of hybrid vehicles with multiple on-board power sources,numerous studies realized by various authors can be found in the literature.Intelligent power management techniques are employed in Refs.[8,22,23].Among them,fuzzy logic bad control methodology has a major position due to its independence of a full mathematical plant model.Another power management method is realized in Ref.[24]for FC/battery
hybrid structure.The mentioned model-bad control in Ref.[24]targets the high fuel economy and sy
stem per-formance utilizing state obrvers.Many other power management techniques take place in the literature and a detailed literature sur-vey on different power management techniques applied to hybrid vehicular systems can be found in Ref.[25].The power manage-
Fig.8.Decomposition of the UDDS power profile using wavelet transform.
