CN103927440A - Method for identifying parameters of equivalent circuit models of lithium batteries - Google Patents

Abstract

The invention discloses a method for identifying parameters of equivalent circuit models of lithium batteries. The method includes construction of objective functions of genetic algorithms, initialization operation, crossover operation, local search and mutation operation. Objective functions are constructed on the basis of average estimation errors. Value ranges of the to-be-identified parameters are set by means of initialization operation. Crossover operation and mutation operation are carried out according to initial probability values. A process for constructing adjacent individuals by the aid of random crossing genetic bit values is adopted during local search. The method has the advantages of low solving time complexity, high global optimization ability, good algorithm convergence performance and few parameter identification errors. The method is suitable for scientific research institutions to verify and apply simulation modeling technologies.

Description

A kind of parameter identification method of lithium battery equivalent-circuit model

Technical field

The present invention relates to a kind of parameter identification method of lithium battery equivalent-circuit model, belong to electric system accumulator system technical field.

Background technology

Lithium-ion-power cell has the advantages such as nominal voltage is high, specific energy is large, the life-span is long, is considered to have in following electric vehicle battery the accumulator of development potentiality, has been widely used in pure electric automobile, hybrid vehicle and fuel cell car.Battery model is the basis of battery charge state (SOC) estimation, performance evaluation, scientific evaluation, efficient management and using, is the tie from outside batteries characteristic to internal state.Accurate dynamic model is significant to electrokinetic cell emulation, optimization and energy management.

Researchist has set up multiple galvanochemistry and the circuit model that can carry out to battery performance description comprehensively at present, but its accuracy depends on parameter identification precision to a great extent.

Summary of the invention

The object of the invention is to, for the simulation study of lithium battery provides basis, and for SOC estimates to provide foundation, the invention provides a kind of parameter identification method of lithium battery equivalent-circuit model.

Technical scheme of the present invention is that the parameter identification method of a kind of lithium battery equivalent-circuit model of the present invention, comprises genetic algorithm objective function, initialization operation, interlace operation, Local Search, mutation operation.

The parameter identification method of a kind of lithium battery equivalent-circuit model of the present invention comprises the following steps:

(1) lithium battery equivalent-circuit model is by open-circuit voltage V _oc, internal resistance R _irwith equivalent capacity C _eccomposition, wherein internal resistance R _ircomprise Ohmage R _orwith polarization resistance R _pr.The k moment is by the electric current I of electric capacity _{ec, k}represent, pass through resistance R _orelectric current I _{or, k}represent battery terminal voltage V _t,krepresent.Equivalent-circuit model represents with following formula:

V _T,k＝V _oc-R _or×I _or,k-R _pr×I _ec,k

$I_{\mathrm{ec},k}=[1-\frac{1-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}}{\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}]\×I_{\mathrm{or},k}+[\frac{1-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}}{\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}]\×I_{\mathrm{or},k-1}+e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}\×I_{\mathrm{ec},k-1}$

Wherein, Δ t represents the time interval between k-1 moment and k moment.

(2) objective function of the equivalent-circuit model structure genetic algorithm based on step (1), represents with following formula:

$\begin{array}{c}\min \left(f\left(O_k^g\right)\right)\\ f\left(O_k^g\right)=\frac{1}{L}\underset{k=1}{\overset{L}{\mathrm{\Σ}}}{(V_{T,k}-{\hat{V}}_{T,k}^g)}^2\end{array}$

Wherein, represent the terminal voltage estimated value of g population.

(3) Population in Genetic Algorithms sizes values g, maximum iteration time m, crossover probability P are set _c, variation probability P _m, V parameter to be identified _oc, R _or, R _prand C _ecspan, and each individuality in initialization population.

(4) according to crossover probability P _cgenerate individual crossover location, then from population, select at random two individualities to carry out interlace operation, until all individualities were all traversed.

(5) to each individuality, generate at random two integers, exchange the gene position of these two round values correspondence positions, and calculate ideal adaptation degree according to step (2) through 3～5 operations, choose the individuality of ideal adaptation degree minimum.

(6) the each gene location to each individuality, generating span is the probable value p of [0,1], when p is less than variation probability P _mtime, replace this gene position numerical value by random number, generate progeny population.

(7) operation of repeating step (5).

(8) calculate the fitness of all individualities if current individual minimum fitness is less than the fitness of current population, the best individuality of population is the individuality of current fitness minimum, otherwise the individuality of current fitness maximum is replaced with to the individuality of current population's fitness maximum.

(9) repeating step (4)～(8), until meet maximum iteration time m, output obtains identified parameters with

The present invention's beneficial effect is compared with the prior art, in individual and overall population, all added Local Search, and Algorithm for Solving time complexity is low, and global optimizing ability is stronger, and algorithm convergence performance is better, and the error of parameter identification is less.The simulation study that the present invention can be lithium battery provides basis, and for SOC estimates to provide foundation, contributes to battery BMS Design and implementation, and the method can be lithium battery model for different scenes simultaneously, as the application such as micro-electrical network, electric automobile provide technical support.

The present invention is applicable to scientific research institution and carries out the checking application of accumulator system Simulation and Modeling Technology.

Embodiment

The specific embodiment of the present invention is as follows:

(1) equivalent-circuit model is by open-circuit voltage V _oc, internal resistance R _irwith equivalent capacity C _eccomposition, wherein internal resistance R _ircomprise Ohmage R _orwith polarization resistance R _pr.The k moment is by the electric current I of electric capacity _{ec, k}represent, pass through resistance R _orelectric current I _{or, k}represent battery terminal voltage V _t,krepresent.Equivalent-circuit model represents with following formula:

V _T,k＝V _oc-R _or×I _or,k-R _pr×I _ec,k

$I_{\mathrm{ec},k}=[1-\frac{1-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}}{\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}]\×I_{\mathrm{or},k}+[\frac{1-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}}{\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}]\×I_{\mathrm{or},k-1}+e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}\×I_{\mathrm{ec},k-1}$

Wherein, Δ t represents the time interval between k-1 moment and k moment.

(2) objective function of the equivalent-circuit model structure genetic algorithm based on step (1), represents with following formula:

$\begin{array}{c}\min \left(f\left(O_k^g\right)\right)\\ f\left(O_k^g\right)=\frac{1}{L}\underset{k=1}{\overset{L}{\mathrm{\Σ}}}{(V_{T,k}-{\hat{V}}_{T,k}^g)}^2\end{array}$

Wherein, represent the terminal voltage estimated value of g population.

(3) Population in Genetic Algorithms sizes values g, maximum iteration time m, crossover probability P are set _c, variation probability P _m, V parameter to be identified _oc, R _or, R _prand C _ecspan, and each individuality in initialization population.

(4) according to crossover probability P _cgenerate individual crossover location, then from population, select at random two individualities to carry out interlace operation, until all individualities were all traversed.

(5) to each individuality, generate at random two integers, exchange the gene position of these two round values correspondence positions, and calculate ideal adaptation degree according to step (2) through 3～5 operations, choose the individuality of ideal adaptation degree minimum.

(6) the each gene location to each individuality, generating span is the probable value p of [0,1], when p is less than variation probability P _mtime, replace this gene position numerical value by random number, generate progeny population.

(7) operation of repeating step (5).

(8) calculate the fitness of all individualities if current individual minimum fitness is less than the fitness of current population, the best individuality of population is the individuality of current fitness minimum, otherwise the individuality of current fitness maximum is replaced with to the individuality of current population's fitness maximum.

(9) repeating step (4)～(8), until meet maximum iteration time m, output obtains identified parameters with

Claims (1)

1. a parameter identification method for lithium battery equivalent-circuit model, is characterized in that, said method comprising the steps of:

(1) equivalent-circuit model is by open-circuit voltage V _oc, internal resistance R _irwith equivalent capacity C _eccomposition, wherein internal resistance R _ircomprise Ohmage R _orwith polarization resistance R _pr; The k moment is by the electric current I of electric capacity _{ec, k}represent, pass through resistance R _orelectric current I _{or, k}represent battery terminal voltage V _t,krepresent; Equivalent-circuit model represents with following formula:

V _T,k＝V _oc-R _or×I _or,k-R _pr×I _ec,k

$I_{\mathrm{ec},k}=[1-\frac{1-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}}{\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}]\×I_{\mathrm{or},k}+[\frac{1-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}}{\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}-e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}]\×I_{\mathrm{or},k-1}+e^{-\mathrm{\Δt}/(C_{\mathrm{ec}}\×R_{\mathrm{pr}})}\×I_{\mathrm{ec},k-1}$

Wherein, Δ t represents the time interval between k-1 moment and k moment;

(2) objective function of the equivalent-circuit model structure genetic algorithm based on step (1), represents with following formula:

$\begin{array}{c}\min \left(f\left(O_k^g\right)\right)\\ f\left(O_k^g\right)=\frac{1}{L}\underset{k=1}{\overset{L}{\mathrm{\Σ}}}{(V_{T,k}-{\hat{V}}_{T,k}^g)}^2\end{array}$

Wherein, represent the terminal voltage estimated value of g population;

(3) Population in Genetic Algorithms sizes values g, maximum iteration time m, crossover probability P are set _c, variation probability P _m, V parameter to be identified _oc, R _or, R _prand C _ecspan, and each individuality in initialization population;

(4) according to crossover probability P _cgenerate individual crossover location, then from population, select at random two individualities to carry out interlace operation, until all individualities were all traversed;

(5) to each individuality, generate at random two integers, exchange the gene position of these two round values correspondence positions, and calculate ideal adaptation degree according to step (2) through 3～5 operations, choose the individuality of ideal adaptation degree minimum;

(6) the each gene location to each individuality, generating span is the probable value p of [0,1], when p is less than variation probability P _mtime, replace this gene position numerical value by random number, generate progeny population;

(7) operation of repeating step (5);

(8) calculate the fitness of all individualities if current individual minimum fitness is less than the fitness of current population, the best individuality of population is the individuality of current fitness minimum, otherwise the individuality of current fitness maximum is replaced with to the individuality of current population's fitness maximum;

(9) repeating step (4)～(8), until meet maximum iteration time m, output obtains identified parameters with