CN106842056A - One kind is based on two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation - Google Patents

Abstract

Two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation is based on the present invention relates to one kind, based on first step intelligent optimization algorithm type, is found in the first optimized variable border and is caused the first optimization object function J₁Maximized first optimized variable value, solves the J corresponding to the first optimized variable value₁Value is the maximum discharge power of subsequent time electrokinetic cell；Based on second step intelligent optimization algorithm type, found in the second optimized variable border and cause the second optimization object function J₂The the second optimized variable value for minimizing, solves the J corresponding to the second optimized variable value₂Value is the minimum charge power of subsequent time electrokinetic cell；The method introduces that two-step intelligent optimization algorithm calculates maximum discharge power and minimum charge power respectively and step is simple, strong applicability, it is easy to on-line implement.

Description

One kind is based on two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation

Technical field

This patent is related to batteries of electric automobile management domain, and more particularly to Prospect of EVS Powered with Batteries peak power is estimated The technical field of meter.

Technical background

The peak power (SoP) of real-time estimation electric automobile power battery can assess electrokinetic cell in different health status (SoH) limit capacity of charge-discharge electric power, Optimized Matching power battery pack are between vehicle dynamic quality and under state-of-charge (SoC) Relation and at utmost play motor regenerating braking energy and reclaim ability.It is accurate to estimate SoP to electric automobile whole control The energy management and optimization of unit have highly important theory significance and practical value, are the weights of cell management system of electric automobile Want function and study hotspot.

Intelligent optimization algorithm type well known in the art mainly includes genetic algorithm, particle cluster algorithm, ant group algorithm, simulation Annealing algorithm etc..More specifically intelligent optimization algorithm is largely being disclosed in the prior art, for example《The sharp intelligent optimizations of Huang You are calculated Method and its application [M] National Defense Industry Press, 2008》With《Li Shiyong, Li Yan intelligent optimization algorithms principle are breathed out with application [M] That shore Polytechnic University Publishing House, 2012》.But current intelligent optimization algorithm is not applied to estimating for electrokinetic cell peak power also Meter.

The method that conventional battery SoP estimates at present, such as neural network, composite pulse algorithm (PNGV HPPC methods), electricity Platen press, SoC methods, do not ensure that the optimality of estimated result, therefore, this patent discloses a kind of based on the optimization of two-step on-line intelligence The electrokinetic cell peak power method of estimation of algorithm.

The content of the invention

For problem above, the invention discloses a kind of electrokinetic cell its peak work based on two-step on-line intelligence optimized algorithm Rate method of estimation.The method introduces intelligent optimization algorithm, and calculates maximum electric discharge respectively using two-step intelligent optimization algorithm Power and minimum charge power, and the calculating of discharge and recharge peak power under each step-length can be completed, step is simple, applicability By force, it is easy to on-line implement.

One kind is based on two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation, including：

Type and the running environment ginseng of step one, initialization first step intelligent optimization algorithm and second step intelligent optimization algorithm Number；

Step 2, defines the first optimized variable, the first optimized variable border and first of first step intelligent optimization algorithm Optimization object function J₁；

Step 3, based on first step intelligent optimization algorithm type, finds in the first optimized variable border and causes the One optimization object function J₁Maximized first optimized variable value, solves the J corresponding to the first optimized variable value₁Value is The maximum discharge power of subsequent time electrokinetic cell；

Step 4, defines the second optimized variable, the second optimized variable border and second of second step intelligent optimization algorithm Optimization object function J₂；

Step 5, based on second step intelligent optimization algorithm type, finds in the second optimized variable border and causes the Two optimization object function J₂The the second optimized variable value for minimizing, solves the J corresponding to the second optimized variable value₂Under value is The minimum charge power of one moment electrokinetic cell；

Step 6, calculates the discharge and recharge peak power of power battery pack.

Preferably, the first optimized variable is the maximum discharge power that electrokinetic cell subsequent time is allowed in the step 2Corresponding maximum discharge current valueFirst optimized variable border is set to [0, I_max], I_maxFor maximum is discharged Electric current.

Preferably, the second optimized variable is the minimum charge power that electrokinetic cell subsequent time is allowed in the step 4Corresponding minimum charge current valueSecond optimized variable border is set to [I_min, 0], I_minFor minimum charges Electric current.

One kind is based on two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation, including

A is walked, and sets the duration parameters Δ T that the battery peak power is estimated；Initialization electrokinetic cell limit work Make the restrictive condition of state, including maximum discharge current and minimum charge current；

B is walked, and gathers and record the current value of moment k electrokinetic cell；

C is walked, and estimates the SoC values of the electrokinetic cell at current time k；

D is walked, and identification obtains moment k electrokinetic cell single order RC equivalent circuit model parameters；

E is walked, and based on above-mentioned equivalent-circuit model, performs the method as described in claims 1 to 3；

F is walked, and into subsequent time k=k+1, is back to b steps.

Preferably, the restrictive condition of the electrokinetic cell limit working condition is put also comprising the upper blanking voltage Umax that charges At least one of lower minimum limitation zmin of blanking voltage Umin, SoC threshold limit value zmax and SoC of electricity.

Preferably, the first optimization object function J₁Expression formula is：

M₁And M₂It is penalty factor, U_minIt is the upper blanking voltage, z of charging_minIt is SoC minimum limit values,For electrokinetic cell exists The k+1 moment exportsWhen terminal voltage,It is electrokinetic cell in the SoC values at k+1 moment.

Preferably, the second optimization object function J₂Expression formula is：

M₃And M₄It is penalty factor, U_maxIt is the upper blanking voltage, z of charging_maxIt is SoC threshold limit values,It is electrokinetic cell Exported at the k+1 momentWhen terminal voltage,It is electrokinetic cell in the SoC values at k+1 moment.

Brief description of the drawings

Fig. 1 is based on the electrokinetic cell peak power method of estimation flow chart of two-step on-line intelligence optimized algorithm；

Fig. 2 single order RC equivalent-circuit model structure charts；

Electrokinetic cell offline parameter identification result in Fig. 3 the present embodiment；

Fig. 4 two-step on line genetic algorithms estimated driving force battery SoP flow charts；

Power battery pack SoP estimated result of Fig. 5 the present embodiment under UDDS operating modes；

The SoP of Fig. 6 the present embodiment each sampled point under UDDS operating modes estimates the used time；

Specific embodiment：

Current symbol is defined as just when being discharged in present invention expression, and rechargeable electric current sign convention is negative.Therefore, discharge Electric current is on the occasion of maximum discharge current is the discharge current of maximum absolute value；Charging current is negative value, therefore minimum charge current It is the charging current of maximum absolute value.Electric discharge peak power is on the occasion of maximum electric discharge peak power is the electric discharge of maximum absolute value Peak power；Charging peaks power is negative value, therefore minimum charging peaks power is the charging peaks power of maximum absolute value.

Electrokinetic cell peak power method of estimation based on two-step on-line intelligence optimized algorithm specifically includes following steps：

A is walked, and sets the duration parameters Δ T that SoP estimates, i.e., estimated SoP values are used to lasting output or input Time span, initialize electrokinetic cell limit working condition restrictive condition.The restrictive condition includes maximum discharge current I_max, minimum charge current I_min, blanking voltage U in charging_max, the lower blanking voltage U of electric discharge_min, SoC threshold limit values z_maxWith SoC most Small limitation z_min。

B is walked, and gathers and record the current value I of moment k electrokinetic cell_k；

C is walked, and estimates the SoC values z of the electrokinetic cell at current time k_k, electrokinetic cell SoC methods of estimation can be ability Method known to domain, such as ampere-hour counting method, the terminal voltage estimation technique, Kalman Filter Estimation method；

D is walked, and identification obtains moment k electrokinetic cell single order RC equivalent circuit model parameters, specifically includes kth moment power The open-circuit voltage U of battery_oc, charging ohmic internal resistance R_chg, electric discharge ohmic internal resistance R_dis, the resistance for reflecting electrokinetic cell polarization phenomenon Hold the resistance R in loop_pAnd timeconstantτ.

The discrimination method of equivalent circuit model parameter can use method well known in the art, and conventional discrimination method can be divided into Off-line identification method and on-line identification method, off-line identification method have genetic algorithm, particle cluster algorithm etc., and on-line identification method has recursion minimum Square law, Kalman Filter Estimation method etc..

E is walked, and based on equivalent-circuit model, performs the SoP values of two-step on-line intelligence optimized algorithm estimated driving force battery, i.e., Obtaining subsequent time electrokinetic cell allows the peak power of lasting Δ T electric dischargesWith the minimum power for chargingIts In, the present invention does not limit intelligent optimization algorithm；

F is walked, and into subsequent time k=k+1, is back to b steps.

Wherein, in above-mentioned e step based on single order RC equivalent-circuit models, perform two-step on-line intelligence optimized algorithm and estimate The SoP values of electrokinetic cell, specifically include following steps：

Two-step on-line intelligence optimized algorithm of the invention includes that first step intelligent optimization algorithm and second step intelligent optimization are calculated Method, wherein the intelligent optimization algorithm type that the first step intelligent optimization algorithm and second step intelligent optimization algorithm are used is by this Art personnel select as needed, the first step intelligent optimization algorithm and second step intelligent optimization algorithm can with identical, Can also be different.Intelligent optimization algorithm type is known in the art technology, including but not limited to genetic algorithm, particle cluster algorithm, Ant group algorithm, simulated annealing etc..

Step one, initializes the running environment parameter of two-step on-line intelligence optimized algorithm, and those skilled in the art are according to true Fixed intelligent optimization algorithm type sets carrying out practically ambient parameter.

For example, if the first step or second step intelligent optimization algorithm type in the step one are genetic algorithm, institute The carrying out practically ambient parameter stated includes population scale, reproductive order of generation, coded system, selection mode, variation mode, intersection side Formula.

Step 2, defines optimized variable, optimized variable border and the optimization object function of first step intelligent optimization algorithm, Take the maximum discharge power of electrokinetic cell subsequent time permissionCorresponding current valueIt is optimized variable, optimization becomes Amount border is set to [0, I_max], I_maxIt is the set maximum discharge current of a steps, the optimization mesh of first step intelligent optimization algorithm Scalar functions J₁Expression formula is as follows：

Wherein, M₁And M₂It is penalty factor, U_minIt is blanking voltage, z in the set charging of a steps_minFor a steps are set The SoC minimum limit values put,Be electrokinetic cell in the SoC values at k+1 moment, if the discrimination method in d steps uses off-line identification Method, based on single order RC equivalent-circuit models,Expression formula it is as follows：

Wherein, Δ t is systematic sampling time interval,For electrokinetic cell is exported at the k+1 momentWhen end electricity Pressure, U_p,kIt is the magnitude of voltage in k moment capacitance-resistances loop,It is that electrokinetic cell is being exportedWhen open-circuit voltage values,It is that electrokinetic cell is being exportedWhen ohm resistance, R_p(z_k) for electrokinetic cell in the capacitance-resistance loop at k moment Resistance value, τ (z_k) it is capacitance-resistance loop time constant of the electrokinetic cell at the k moment, I_kIt is electrokinetic cell in the current value at k moment, z_k It is electrokinetic cell in the SoC values at k moment.If the discrimination method in d steps uses on-line identification method, based on single order RC equivalent circuits Model,Expression formula it is as follows：

Wherein, U_oc(z_k) it is open-circuit voltage values of the electrokinetic cell at the k moment, R_dis(z_k) it is the ohm at electrokinetic cell k moment Resistance value；

Expression formula it is as follows：

Wherein, η_dIt is electric discharge coulombic efficiency, C is the current capacities of electrokinetic cell；

Step 3, first step intelligent optimization algorithm is performed according to step one and two setting, i.e., set according to step one The running environment parameter of algorithm, the first step intelligent optimization algorithm type determined using step one finds [0, I_max] make in region Obtain optimization object function J₁It is maximizedSolveThe corresponding J of value₁Value is subsequent time electrokinetic cell Electric discharge SoP values

Step 4, defines optimized variable, optimized variable border and the optimization object function of second step intelligent optimization algorithm, Take the minimum charge power of electrokinetic cell subsequent time permissionCorresponding current valueIt is optimized variable, optimization becomes Amount border is set to [I_min, 0], I_minIt is the set minimum charge current of a steps, the optimization mesh of second step intelligent optimization algorithm Scalar functions J₂Expression formula is as follows：

Wherein, M₃And M₄It is penalty factor, U_maxIt is blanking voltage, z in the set charging of a steps_maxFor a steps are set The SoC threshold limit values put,It is electrokinetic cell in the SoC values at k+1 moment；If the discrimination method in d steps uses off-line identification Method, based on single order RC equivalent-circuit models,Expression formula it is as follows：

Wherein,For electrokinetic cell is exported at the k+1 momentWhen terminal voltage,It is electrokinetic cell defeated Go outWhen open-circuit voltage values,It is that electrokinetic cell is being exportedWhen ohm resistance, if d step in Discrimination method uses on-line identification method, based on single order RC equivalent-circuit models,Expression formula it is as follows：

Wherein, U_oc(z_k) it is open-circuit voltage values of the electrokinetic cell at the k moment, R_chg(z_k) it is the ohm at electrokinetic cell k moment Resistance value；

Expression formula it is as follows：

Wherein, η_cIt is charging coulombic efficiency；

Step 5, second step intelligent optimization algorithm is performed according to step one and four setting, i.e., set according to step one Algorithm design parameter, the second step intelligent optimization algorithm type determined using step one finds [I_min, 0] and optimization is caused in region Object function J₂MinimizeSolveThe corresponding J of value₂Value is the charging SoP of subsequent time electrokinetic cell Value

Step 6, calculates the discharge and recharge SoP values of power battery packWithFormula is as follows；

Wherein, n_sIt is the number of modules that power battery pack is connected in series, n_pIt is the electrokinetic cell being connected in parallel in each module Monomer number；With reference to experiment, the present invention is described in detail, using the electrokinetic cell of certain 18650e type in embodiment, this Embodiment does not constitute restriction to the present invention.

Step 1：The duration parameters Δ T=1s that SoP estimates is set, the limit of electrokinetic cell limit working condition is initialized Condition processed is shown in Table 1：

The restrictive condition parameter list of the electrokinetic cell limit working condition of table 1

Step 2, gathers and records the current value I of moment k electrokinetic cell_k；

Step 3, estimates the SoC values z of current time k electrokinetic cell_k, the present embodiment is using ampere-hour counting method estimated driving force electricity Pond SoC, formula is as follows：

Wherein, Δ t is sampling time interval, in the present embodiment, Δ t=1s, C=1.299Ah, η_d=1, η_c=0.98；

Step 4, identification obtains moment k electrokinetic cell equivalent circuit model parameter θ_k, the present embodiment is using off-line identification side Method, carries out Metro cycle (UDDS) working condition measurement to electrokinetic cell first, and the SoC codomains in working condition measurement will be tested Electric current and terminal voltage data be divided into 13 minizones, then based on single order RC equivalent-circuit models with terminal voltage root-mean-square error Minimum principle carries out off-line identification, finally using formula respectively to U_oc、R_chg、R_dis, R_pIt is fitted with τ, fitting formula is as follows：

Wherein, λ_xyThe coefficient (1≤x≤6,1≤y≤5, the equal round numbers of x and y) of parameter fitting formula is, step 3 is estimated The z of meter_kSubstituting into formula (9) can obtain k electrokinetic cell equivalent circuit model parameters, and the off-line identification result of the present embodiment is shown in figure 3；

Step 5, based on single order RC equivalent-circuit models, performs two-step on-line intelligence optimized algorithm estimated driving force battery SoP valuesWithThe first step and second step intelligent optimization algorithm of the present embodiment all use genetic algorithm；

Step 6, into subsequent time k=k+1, is back to the 2nd step.

In the step 5 based on single order RC equivalent-circuit models, perform two-step on-line intelligence optimized algorithm estimated driving force The SoP values of battery, as shown in figure 4, specifically including following steps：

Step 101, initializes the calculating parameter of two-step on line genetic algorithms, and the setting of design parameter is shown in Table 2：

The two-step genetic algorithm design parameter of table 2 sets table

Step 102, define first step genetic algorithm optimized variable beOptimized variable border is set to [0,27], By J₁Optimization object function is defined as, J is taken here₁It is the fitness function in genetic algorithm, in the present embodiment, M₁Take 10⁶, M₂Take 10⁷；

Step 103, first step genetic algorithm is performed according to step 101 and 102 setting, and the tool shown in table 2 is set first Body parameter, is encoded, [0, I according to real coding mode_max] 50 individualities of generation and to calculate each individual at random in region Fitness function value, then wheel disc selection, heuristic intersection and uniform variation are carried out to individuality, if current reproductive order of generation is not up to 100, the step of calculating each individual adaptation degree functional value is returned to, if current reproductive order of generation reaches 100, export current population In optimal fitness function value, be after real number decoding the electric discharge SoP values of subsequent time electrokinetic cellThis reality The SoP for applying example estimates that environment selects UDDS operating modes；

Step 104, define second step genetic algorithm optimized variable beOptimized variable border be set to [- 27, 0], by J₂Optimization object function is defined as, because the fitness function of genetic algorithm must have nonnegativity, and fitness letter Numerical value is bigger, and the individuality that represents is more outstanding, therefore takes-J here₂It is the fitness function in genetic algorithm, in the present embodiment, M₃Take 10⁶, M₄Take 10⁷；

Step 105, second step genetic algorithm is performed according to step 101 and 104 setting, and the tool shown in table 2 is set first Body parameter, is encoded, in [I according to real coding mode_min, 0] and 50 individualities of generation and to calculate each individual at random in region Fitness function value, then wheel disc selection, heuristic intersection and uniform variation are carried out to individuality, if current reproductive order of generation is not up to 100, the step of calculating each individual adaptation degree functional value is returned to, if current reproductive order of generation reaches 100, export current population In optimal fitness function value, take after real number decoding the charging SoP values of opposite number as subsequent time electrokinetic cell

Step 106, the discharge and recharge SoP values of power battery pack are calculated according to formula (9)WithThe present embodiment Middle n_sIt is 1, n_pIt is 20；

Repeat step 6, k completes the present invention and implements dynamic under UDDS operating modes from 1 finish time for being added to UDDS operating modes The checking that power battery set charge/discharge SoP estimates, estimated result is shown in Fig. 5.SoP estimates that computing uses Intel i7 in the present embodiment 2.20GHz processors and 8GB internal memories, the SoP on each sampled point under UDDS operating modes estimate that the used time sees Fig. 6.

It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, Er Qie In the case of without departing substantially from spirit or essential attributes of the invention, the present invention can be in other specific forms realized.Therefore, no matter From the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, the scope of the present invention is by appended power Profit requires to be limited rather than described above, it is intended that all in the implication and scope of the equivalency of claim by falling Change is included in the present invention.Any reference in claim should not be considered as the claim involved by limitation.

Claims (7)

1. It is 1. a kind of to be based on two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation, it is characterised in that：

   The type and running environment parameter of step one, initialization first step intelligent optimization algorithm and second step intelligent optimization algorithm；

   Step 2, defines the first optimized variable, the first optimized variable border and first optimization of first step intelligent optimization algorithm Object function J₁；

   Step 3, based on first step intelligent optimization algorithm type, finds in the first optimized variable border and causes that first is excellent Change object function J₁Maximized first optimized variable value, solves the J corresponding to the first optimized variable value₁Value is as next The maximum discharge power of moment electrokinetic cell；

   Step 4, defines the second optimized variable, the second optimized variable border and second optimization of second step intelligent optimization algorithm Object function J₂；

   Step 5, based on second step intelligent optimization algorithm type, finds in the second optimized variable border and causes that second is excellent Change object function J₂The the second optimized variable value for minimizing, solves the J corresponding to the second optimized variable value₂Value i.e. lower a period of time Carve the minimum charge power of electrokinetic cell；

   Step 6, calculates the discharge and recharge peak power of power battery pack.
2. 2. the method for claim 1, it is characterised in that：The first optimized variable is next electrokinetic cell in the step 2 The maximum discharge power that moment allowsCorresponding maximum discharge current valueFirst optimized variable border is set to [0,I_max], I_maxIt is maximum discharge current.
3. 3. the method for claim 1, it is characterised in that：The second optimized variable is next electrokinetic cell in the step 4 The minimum charge power that moment allowsCorresponding minimum charge current valueSecond optimized variable border is set to [I_min, 0], I_minIt is minimum charge current.
4. It is 4. a kind of to be based on two-step on-line intelligence optimized algorithm electrokinetic cell peak power method of estimation, it is characterised in that：

   A is walked, and sets the duration parameters Δ T that the battery peak power is estimated；Initialization electrokinetic cell limit work shape The restrictive condition of state, including maximum discharge current and minimum charge current；

   B is walked, and gathers and record the current value of moment k electrokinetic cell；

   C is walked, and estimates the SoC values of the electrokinetic cell at current time k；

   D is walked, and identification obtains moment k electrokinetic cell single order RC equivalent circuit model parameters；

   E is walked, and based on above-mentioned equivalent-circuit model, performs the method as described in claims 1 to 3；

   F is walked, and into subsequent time k=k+1, is back to b steps.
5. 5. method as claimed in claim 4, it is characterised in that：The restrictive condition of the electrokinetic cell limit working condition is also wrapped Containing the upper blanking voltage U that charges_max, the lower blanking voltage U of electric discharge_min, SoC threshold limit values z_maxLimitation z minimum with SoC_minIn at least One.
6. 6. the method as described in claim any one of 2-4, it is characterised in that the first optimization object function J₁Expression formula is：

   $J_1=I_{\max ,k+1}^{dis}{U}_{t,k+1}^{dis}+M_{1}min(U_{t,k+1}^{dis}-U_{\min },0)+M_{2}min(z_{k+1}^{dis}-z_{\min },0)$

   M₁And M₂It is penalty factor, U_minIt is electric discharge lower blanking voltage, z_minIt is SoC minimum limit values,It is electrokinetic cell in k+1 Moment exportsWhen terminal voltage,It is electrokinetic cell in the SoC values at k+1 moment.
7. 7. the method as described in claim any one of 2-4, it is characterised in that the second optimization object function J₂Expression formula is：

   $J_2=I_{min,k+1}^{chg}{U}_{t,k+1}^{chg}+M_{3}max(U_{t,k+1}^{chg}-U_{max},0)+M_{4}max(z_{k+1}^{chg}-z_{max},0)$

   M₃And M₄It is penalty factor, U_maxIt is the upper blanking voltage, z of charging_maxIt is SoC threshold limit values,It is electrokinetic cell in k+1 Moment exportsWhen terminal voltage,It is electrokinetic cell in the SoC values at k+1 moment.