CN105891727A - Method and system for estimating state of charge of dual-variable structured filtering power battery - Google Patents

Abstract

The invention discloses a method and a system for estimating the state of charge (SOC) of a dual-variable structured filter power battery. The method comprises the following steps: firstly, establishing a battery equivalent model, identifying parameters of the power battery model by adopting first variable structure filtering, fitting the relationship of open circuit voltage (OCV) and the SOC of the power battery, and furthermore estimating the SOC by adopting second variable structure filtering. A variable structure filtering parameter beta beta value has relatively large influence on modification increment, and beta beta is adjusted by introducing fuzzy rules in a self-adaptive manner. A voltage and current sensor of the estimation system is mounted on a power battery to be detected, a computing module for implementing the method is arranged inside a microprocessor, and the microprocessor is used for displaying the SOC value which is estimated at present on line, and can be connected with a CAN controller of an automobile. By adopting the method and the system, the parameters of the power battery can be identified on line; the modification increment of the variable structure filtering parameters can be adjusted in a self-adaptive fuzzy manner, the method is concise, small in calculating amount, easy to achieve, high in precision and small in dependency on SOC initial values.

Description

The method of estimation of the power battery charged state of a kind of pair of structure changes filtering and system

Technical field

The present invention relates to the electrokinetic cell state estimation field of electric automobile, the power of a kind of pair of structure changes filtering The method of estimation of battery charge state and system.

Background technology

New-energy automobile is a kind of effective ways solving orthodox car environmental pollution, and electrokinetic cell is new-energy automobile One of core key technology.The estimated value of the state-of-charge (State of Charge is called for short SOC) of electrokinetic cell is in high precision Important parameter in power vehicle driving process, is also the important evidence of electric automobile during traveling state.

Now commonly use power battery charged state evaluation method to have: ampere-hour integration method, open-circuit voltage method, Kalman filtering method and Particle filter method etc..Time actually used, these methods all have the defect being difficult to overcome.Ampere-hour integration method is by onboard sensor precision Restriction, be easily caused cumulative errors, the state-of-charge SOC estimation error causing electrokinetic cell is the biggest.Opening of open-circuit voltage method The collection of road voltage data needs battery to stand for a long time, could accurately measure, thus be not suitable in actual driving conditions SOC On-line Estimation.Kalman filtering method is used for the state-of-charge SOC of linear electrokinetic cell and estimates.EKF The state-of-charge SOC of electrokinetic cell is set up state-space model as a state variable by method, Unscented kalman filtering method etc., Obtained the state-of-charge SOC minimum variance estimate of electrokinetic cell by recursion, but these Kalman filtering methods improved are to mould The dependency of type is the highest；Particle filter method need with substantial amounts of sample size could the posterior probability of approximation system effectively close Degree, operand is the biggest.

The state-of-charge SOC estimation method of the most existing electrokinetic cell due to deviation accumulation, electrokinetic cell non-linear Deng affecting estimation precision, the most not yet there are SOC estimation on line precision height, convergence rate very fast, and low to initial value precision dependency degree The state-of-charge SOC method of estimation of electrokinetic cell.

Summary of the invention

It is an object of the invention to design the method for estimation of a kind of power battery charged state using double structure changes filtering, adopt Filter battery model parameter identification with a structure changes, use another fuzzy variable structure to filter, battery model parameter is repaiied Positive increment carries out adaptive fuzzy adjustment, and it is effective that this method not only ensures that SOC estimated value is restrained, and ensures the essence of estimated value Degree.Double structure changes filter structures are succinct, adapt to SOC difference just state of value, it is achieved the high accuracy of SOC is estimated.

The method of estimation that it is another object of the present invention to the power battery charged state according to a kind of pair of structure changes filtering sets The estimating system of the power battery charged state of meter double structure changes filtering, uses structure changes to filter battery model parameter identification, The estimation of SOC is realized through having the fuzzy variable structure filtering of the self-adaptative adjustment revising increment.

The method of estimation key step of the power battery charged state SOC of a kind of pair of structure changes filtering of present invention design is such as Under:

I, the parameter identification of electrokinetic cell

I-1, electrokinetic cell discrete model

The present invention uses most widely used Thevenin model to be battery equivalent model, describe battery static state and Dynamic property.The polarization resistance R of battery_pPolarization capacity C with battery_pComposition single order reinforced concrete structure in parallel, represents that the polarization of battery is anti- Should, RC both end voltage is U_pT () represents battery terminal voltage；Concatenation ohmage R₀, flow through ohmic internal resistance R₀Electric current be i (t), Uoc (t) is the open-circuit voltage OCV (Open circuit voltage) of battery, and sampling obtains battery terminal voltage U (t) and flows through Ohmic internal resistance R₀Electric current i (t).

Described battery valid model mathematic(al) representation is as follows:

$\left\{\begin{array}{c}\frac{{\mathrm{dU}}_p\left(1\right)}{dt}=-\frac{U_p\left(t\right)}{R_p{C}_p}+\frac{i\left(t\right)}{C_p}\\ U\left(t\right)=U_{OC}\left(t\right)-R_{0}i\left(1\right)-U_p\left(t\right)\end{array}\right.---\left(1\right)$

Use the backward difference alternative approach battery model discretization to formula (1):

U_k-U_oc,k=a (U_k-1-U_oc,k-1)+bI_k+cI_k-1 (2)

Wherein k is current time, and k-1 is previous moment, and a, b and c are respectively model parameter.

Charge and discharge process in view of electrokinetic cell is slow, and open-circuit voltage Uoc change is less, i.e. U_oc,k=U_oc,k-1, with This, formula (2) is arranged the discrete model of electrokinetic cell equivalent model is:

U_k=a_kU_k-1+b_kI_k+c_kI_k-1+(1-a_k)U_oc,k (3)

Wherein, a_k、b_k、c_kAs follows with battery backward difference model parameter relation:

$R_0=\frac{c_k}{a_k},$

$R_p=\frac{-a_k{b}_k-c_k}{a_k(1-a_k)},$

$C_p=\frac{{{\mathrm{Ta}}_k}^2}{-a_k{b}_k-c_k},$

In formula, T is the sampling period, and T is 0.5 second to 2 seconds.

By the parameter identification of electrokinetic cell discrete model, obtain electrokinetic cell parameter open-circuit voltage Uoc, ohm of battery Internal resistance R₀, polarization resistance Rp and the polarization capacity Cp of battery.

The parameter identification of the electrokinetic cell model of I-2 first structure changes filtering

Use change the first structure filtering that the discrete model of battery equivalent model is carried out parameter identification.Select according to formula (3) Select systematic state variable X_k=[a_k,b_k,c_k,(1-a)U_oc,k]^T, obtaining state equation is:

$\left\{\begin{array}{c}{X}_k=X_{k-1}+w_{k-1}\\ Z_k=U_k={\mathrm{Cx}}_k+v_k\end{array}\right.---\left(4\right)$

Wherein k is current time, x_k∈R_n× 1 is system mode vector；z_k∈R_m× 1 is measuring state variable；w_kAnd v_k It is respectively system zero average stochastic process noise and measures noise.Mainly caused by sensor accuracy, model error etc.；C is to survey Amount equation coefficient: C=[U_k-1,I_k,I_k-1,1]。

Parameter identification process under structure changes filtering-I algorithm is as follows:

$\left[\begin{array}{c}{\hat{z}}_{k|k-1}\\ {\hat{z}}_{k|k}\end{array}\right]=C\left[\begin{array}{c}{\hat{X}}_{k|k-1}\\ {\hat{X}}_{k|k}\end{array}\right]---\left(5\right)$

${\hat{X}}_{k|k-1}={\hat{X}}_{k-1|k-1}---\left(6\right)$

${\hat{X}}_{k|k}={\hat{X}}_{k|k-1}+K_k---\left(7\right)$

First structure changes filteringIt is the systematic state variable X correction value in the k-1 moment,It it is system mode Variable X in the correction value in k moment,It it is the systematic state variable X predictive value in the k moment；It it is measuring state variable Z is at the predictive value in k moment；It it is the measuring state variable z correction value in the k moment；K_kIt it is the parameter identification correction in the k moment Increment, i.e. systematic state variable X are at the predictive value in k momentCorrection.

First structure changes filtering is by adjusting the correction increment K of structure changes filtering_k, constantly revise state variable X in the k moment Predictive valueThe correction increment K in k moment_kFor:

$\left\{\begin{array}{c}{e}_{z_{k|k-1}}=z_k-{\hat{z}}_{k|k-1}\\ e_{z_{k-1|k-1}}=z_{k-1}-{\hat{z}}_{k-1|k-1}\end{array}\right.---\left(9\right)$

(8), z in (9) formula_kThe measured value in terminal voltage k moment when being battery operation, i.e. U_k；It is to survey in the current k moment Error between actual value and the k moment predictive value of amount state variable z；It is the true of k-1 moment measuring state variable Error between value and k-1 moment predictive value；C^-1It it is the inverse matrix of vector C；β, γ are constant values, and span is 0～1, And the value of β directly affects the accuracy of structure changes filtering parameter identification；ο is Schur product, i.e. two matrix corresponding element phases The result taken advantage of；Sat is saturation function, and wherein Ψ is the smooth boundary layer thickness of the first structure changes filtering, the first structure changes filtering The vector definition of sat saturation function as follows:

$sat(e_{z_{k|k-1}},\mathrm{\ψ})={\left[\begin{array}{ccc}sat(e_{z_{1,k|k-1}},{\mathrm{\ψ}}_1)& \mathrm{...}& sat(e_{z_{n,k|k-1}},{\mathrm{\ψ}}_n)\end{array}\right]}^T---\left(10\right)$

Wherein the saturation function sat of the first structure changes filtering is defined as follows:

$sat(e_{z_{i,k|k-1}},{\mathrm{\ψ}}_i)=\left\{\begin{array}{cc}{e}_{z_{i,k|k-1}}/{\mathrm{\ψ}}_i& e_{z_{i,k|k-1}}\≤{\mathrm{\ψ}}_i\\ sign\left(e_{z_{i,k|k-1}}\right)& e_{z_{i,k|k-1}}>{\mathrm{\ψ}}_i\end{array}\right.---\left(11\right)$

Wherein Ψ_iIt is for deviationThe border be given to introduce boundary region, | Ψ_i| for the thickness of boundary region, It is taken as constant, | Ψ_i|=0.01～0.03；Sign represents sign function:

$sign\left(e_{z_{i,k|k-1}}\right)=\left\{\begin{array}{c}{e}_{z_{i,k|k-1}}>0\\ e_{z_{i,k|k-1}}=0\\ e_{z_{i,k|k-1}}<0\end{array}\right..$

Filtered by above first structure changes, obtain system mode vector X_k=[a_k,b_k,c_k,(1-a)U_oc,k]^TEstimation Value isThe parameter value of electrokinetic cell model is obtained: ohmic internal resistance R from system mode vector estimated value₀, polarization resistance Rp, pole Change electric capacity Cp and estimated value U of k moment open-circuit voltage_{Oc, k}, i.e. OCV.

II, the matching of electrokinetic cell open-circuit voltage OCV-SOC relation

In open-circuit voltage algorithm, have good consistent with SOC value with the open-circuit voltage OCV of vehicle mounted dynamic battery of type Property, the high price approximation by polynomi-als fit mathematics relationship model of OCV-SOC is as follows:

$\begin{array}{c}{U}_{oc,k}=g\left({\mathrm{SOC}}_k\right)\\ =h_1{{\mathrm{SOC}}_k}^8+h_2{{\mathrm{SOC}}_k}^7+h_3{{\mathrm{SOC}}_k}^6+h_4{{\mathrm{SOC}}_k}^5+h_5{{\mathrm{SOC}}_k}^4\\ +h_6{{\mathrm{SOC}}_k}^3+h_7{{\mathrm{SOC}}_k}^2+h_8{{\mathrm{SOC}}_k}^1+h_9\end{array}---\left(12\right)$

In formula: h₁～h₉For the coefficient under OCV-SOC high price fitting of a polynomial, obtain after approaching matching: h₁=2.10 × 10³, h₂=-7.38 × 10³, h₃=9.98 × 10³, h₄=-6.23 × 10³, h₅=1.40 × 10³, h₆=3.26 × 10², h₇=-2.40 ×10², h₈=47.98, h₉=22.27.SOC_kRepresent under DST (Dynamic StressTest ambulatory stress test) operating mode Use high precision electro flow measurement, by generally acknowledged SOC definition method obtain in k moment battery dump energy value.

III, the SOC estimation method of the second structure changes filtering

Use the second structure changes filtering to carry out the estimation on line of SOC, the convergence of SOC estimated value can be effectively ensured.Select Polarization capacity C in SOC and battery model_pTerminal voltage U_p,kAs the system state variables of the second structure changes filtering, i.e. XX_k= [SOC_k U_p,k]^T, the state equation of system and measurement equation are as follows:

$\left\{\begin{array}{c}{\mathrm{XX}}_k={\mathrm{AXX}}_{k-1}+{\mathrm{BI}}_{k-1}+{\mathrm{ww}}_k\\ {\mathrm{ZZ}}_k=U_k=U_{oc,k}-R_0{I}_k-U_{P,k}+{\mathrm{vv}}_k\end{array}\right.---\left(13\right)$

Wherein:

(13) in formula: T is the sampling period；Q_NFor battery rated capacity；η is discharge and recharge coulombic efficiency；

R_pRepresent the polarization resistance of battery；C_pRepresent the polarization capacity of battery；U_oc,kRepresent the open-circuit voltage of k moment battery； R_oRepresent the ohmic internal resistance of battery；I_kRepresent that the k moment flows through ohmic internal resistance R_oElectric current；U_kRepresent k moment end during battery operation Voltage；zz_kIt is the measuring state variable of the second structure changes filtering, ww_kAnd vv_kThe system zero being respectively the second structure changes filtering is equal Value stochastic process noise and measurement noise, the system zero average stochastic process noise w that its variance filters with the first structure changes_kAnd survey Amount noise v_kVariance different.Mainly caused by sensor accuracy, model error etc..

Formula (12) is substituted into measurement equation (13) arrange:

$\begin{array}{c}{U}_k={\mathrm{CCX}}_k+h_1{{\mathrm{SOC}}_k}^8+h_2{{\mathrm{SOC}}_k}^7+h_3{{\mathrm{SOC}}_k}^6+h_4{{\mathrm{SOC}}_k}^5\\ +h_5{{\mathrm{SOC}}_k}^4+h_6{{\mathrm{SOC}}_k}^3+h_7{{\mathrm{SOC}}_k}^2+h_9-R_0{I}_k\end{array}---\left(14\right)$

Wherein CC is the coefficient measuring equation, CC=[h₈-1]；SOC_kRepresent the dump energy estimated value of k moment battery.

Obtain according to the first structure changes Filtering Formula (7) in I-2 step:

When state updatesIn KK_kFor revising increment, i.e.Correction value:

Wherein:It is the system state variables xx correction value in the k moment of the second structure changes filtering,It is The systematic state variable X X of two structure changes filtering is at the predictive value in k moment；CC^-1It it is the converse matrix of vector CC；

$\left\{\begin{array}{c}{e}_{{\mathrm{zz}}_{k/k-1}}=Z{Z}_k-Z{\hat{Z}}_{k/k-1}\\ e_{{\mathrm{zz}}_{k-1/k-1}}={\mathrm{ZZ}}_{k-1}-Z{\hat{Z}}_{k-1/k-1}\end{array}\right.$

It it is the mistake k moment that the second structure changes filters measuring between actual value and the predictive value of system state variables Difference；Be k-1 constantly of the second structure changes filtering measures the actual value of system state variables and revised predictive value it Between error；ο is Schur product；Sat is saturation function, and wherein ψ ψ is the smooth boundary layer thickness of the second structure changes filtering, the The vector of the sat saturation function of two structure changes filtering is defined as follows:

$sat(e_{{\mathrm{zz}}_{k|k-1}},\mathrm{\&psi;}\mathrm{\&psi;})={\left[\begin{array}{ccc}sat(e_{{\mathrm{zz}}_{1,k|k-1}},{\mathrm{\&psi;\&psi;}}_1)& \mathrm{...}& sat(e_{{\mathrm{zz}}_{n,k|k-1}},{\mathrm{\&psi;\&psi;}}_n)\end{array}\right]}^T,$

Wherein the saturation function sat of the second structure changes filtering is defined as follows:

$sat(e_{{\mathrm{zz}}_{i,k|k-1}},{\mathrm{\&psi;\&psi;}}_i)=\left\{\begin{array}{cc}{e}_{{\mathrm{zz}}_{i,k|k-1}}/{\mathrm{\&psi;\&psi;}}_i& e_{{\mathrm{zz}}_{i,k|k-1}}\&le;{\mathrm{\&psi;\&psi;}}_i\\ sign\left(e_{{\mathrm{zz}}_{i,k|k-1}}\right)& e_{{\mathrm{zz}}_{i,k|k-1}}>{\mathrm{\&psi;\&psi;}}_i\end{array}\right.,$

Wherein ψ ψ_iIt is for deviationThe border introducing boundary region and provide, | ψ ψ_i| for the thickness of boundary region, it is taken as Constant, value 0.01～0.03；Sign represents sign function, and its rule is as follows:

$sign\left(e_{{\mathrm{zz}}_{i,k|k-1}}\right)=\left\{\begin{array}{c}{e}_{{\mathrm{zz}}_{i,k|k-1}}>0\\ e_{{\mathrm{zz}}_{i,k|k-1}}=0\\ e_{{\mathrm{zz}}_{i,k|k-1}}<0\end{array}\right..$

(15) in the correction incremental computations formula of formula, β β be second structure changes filtering terminal voltage estimated value and terminal voltage true Error between measured valueCoefficient, γ γ be second structure changes filtering terminal voltage correction value and terminal voltage truly survey Error between valueCoefficient.

Lot of experiments finds, β β is to revising increment KK_kValue impact is relatively big and controllability is strong, and the inventive method is mainly to ginseng Number β β value is adjusted.

In order to suppression system is in the buffeting of converged state, the present invention also introduces fuzzy rule, is receiving for structure changes filtering It is easily generated vibration after holding back, causes estimating the characteristic that result produces bigger error, when structure changes filtering carries out SOC estimation on line, Introduce fuzzy rule and adjust structure changes filtering parameter β β, strengthen the adaptability revising increment.Choose succinctly and effectively, both guaranteed SOC precision ensures again the β β value of SOC convergence rate, and the fuzzy rule of β β value is as follows:

$\mathrm{\&beta;}\mathrm{\&beta;}=\left\{\begin{array}{cc}1& e_{{\mathrm{zz}}_{k/k-1}}\&GreaterEqual;0.034U_N\\ 0.1\&CenterDot;e_{{\mathrm{zz}}_{k/k-1}}& 0.034U_N>e_{{\mathrm{zz}}_{k/k-1}}>-0.034U_N\\ 1& e_{{\mathrm{zz}}_{k/k-1}}\&le;-0.034U_N\end{array}\right.---\left(16\right)$

In formulaIt it is the error between actual value and the predictive value measuring system state variables；U_NSpecified electricity for battery Pressure, the present invention takes 12～144V.

Adaptive fuzzy structure changes filtering being carried out parameter by formula (16) adjusts, and is fuzzy variable structure filtering.Should Fuzzy variable structure filtering and the structure changes for battery parameter identification filter to combine and constitute double structure changes filter of the present invention The SOC method of estimation of ripple, tries to achieve the SOC value of electrokinetic cell.

The present invention is according to double changes of the method for estimation design of the power battery charged state of above-mentioned a kind of pair of structure changes filtering The estimating system of power battery charged state SOC of structure filtering, including microprocessor, analog-to-digital conversion module, current sensor, Voltage sensor.

Voltage sensor and current sensor are respectively arranged in electrokinetic cell port to be detected, detection electrokinetic cell end electricity Pressure and the electric current of port.Voltage, current sensor connect microprocessor, microprocessor output present battery through analog-to-digital conversion module The estimated value of state-of-charge.Microprocessor contains data storage and program storage, and described program storage is contained within becoming knot Structure filtering electrokinetic cell parameter identification module, open-circuit voltage and SOC relation fitting module and parameter adaptive fuzzy adjustment mould Block.In the described data storage storage voltage of electrokinetic cell, current detecting data and program storage, each module calculates process The data of middle generation.

Microprocessor is connected with display screen, the estimated value of the battery charge state that Real time displaying is current, i.e. current estimation SOC value.

Microprocessor is furnished with CAN interface, in order to be connected with the CAN controller of automobile, i.e. with other electrical communications of automobile even Connect.

Voltage sensor and current sensor detection obtain the simulation letter of the electric current of current power battery terminal voltage and port Number through analog-to-digital conversion module be converted to correspondence digital signal, send into microprocessor, in microprocessor structure changes filtering power Battery parameter identification module, open-circuit voltage and SOC relation fitting module and parameter adaptive fuzzy adjustment module, according to currently Terminal voltage and current value, the method for estimation of power battery charged state filtered by double structure changes of the present invention estimates, Obtain the estimated value of current battery charge state.

Compared with prior art, the method for estimation of the power battery charged state of a kind of pair of structure changes of present invention filtering be System advantage is: 1, the system of structure changes filtering is easily achieved, simple for structure, can on-line identification electrokinetic cell parameter；2, to becoming knot Structure filtering carries out parameters revision increment and carries out adaptive fuzzy adjustment, introduces fuzzy control rule, obtains double structure changes filtering SOC method of estimation；3, this method of estimation is succinct, and operand is little, and estimated accuracy is high, little to SOC initial value dependency；Knot is become with routine The SOC of structure filtering estimates to compare, and the estimated accuracy of SOC brings up to 1.50% from original 11.09%.

Accompanying drawing explanation

Fig. 1 is the electricity described in step I-1 of the method for estimation embodiment of the power battery charged state of double structure changes filtering Pond equivalent model schematic diagram；

Fig. 2 be the method for estimation embodiment of the power battery charged state of double structure changes filtering fuzzy variable structure filtering with The SOC method of estimation schematic diagram filtered for the structure changes of battery parameter identification；

Fig. 3 is that the SOC of the method for estimation embodiment of the power battery charged state of double structure changes filtering estimates flow process signal Figure；

Fig. 4 is the structural representation of the estimating system embodiment of the power battery charged state of double structure changes filtering.

Detailed description of the invention

The method of estimation embodiment of the power battery charged state SOC of double structure changes filtering

The flow chart of the method for estimation embodiment of the power battery charged state SOC of this pair of structure changes filtering as it is shown on figure 3, Key step is as follows:

I, the parameter identification of electrokinetic cell

I-1, electrokinetic cell discrete model

This example uses Thevenin model to be battery equivalent model, describes static state and the dynamic property of battery, such as Fig. 1 institute Show, the polarization resistance R of battery_pPolarization capacity C with battery_pComposition single order reinforced concrete structure in parallel, represents the polarization reaction of battery, RC Both end voltage is U_pT () represents battery terminal voltage；Concatenation ohmage R₀, flow through ohmic internal resistance R₀Electric current be i (t), Uoc (t) For the open-circuit voltage OCV of battery, sampling obtains battery terminal voltage U (t) and flows through ohmic internal resistance R₀Electric current i (t).

This battery equivalent model mathematic(al) representation is as follows:

$\left\{\begin{array}{c}\frac{{\mathrm{dU}}_p\left(t\right)}{dt}=-\frac{U_p\left(t\right)}{R_p{C}_p}+\frac{i\left(t\right)}{C_p}\\ U\left(t\right)=U_{OC}\left(t\right)-R_{0}i\left(t\right)-U_p\left(t\right)\end{array}\right.---\left(1\right)$

Use the backward difference alternative approach battery model discretization to formula (1):

U_k-U_oc,k=a (U_k-1-U_oc,k-1)+bI_k+cI_k-1 (2)

Wherein k is current time, and k-1 is previous moment, and a, b and c are respectively model parameter.

Formula (2) is arranged the discrete model of electrokinetic cell model is:

U_k=a_kU_k-1+b_kI_k+c_kI_k-1+(1-a_k)U_oc,k (3)

Wherein, a_k、b_k、c_kAs follows with battery backward difference model parameter relation:

$R_0=\frac{c_k}{a_k},$

$R_p=\frac{-a_k{b}_k-c_k}{a_k(1-a_k)},$

$C_p=\frac{{{\mathrm{Ta}}_k}^2}{-a_k{b}_k-c_k},$

In formula, T is the sampling period, and this example T is 1 second.

By the parameter identification of electrokinetic cell discrete model, obtain electrokinetic cell parameter open-circuit voltage Uoc, ohm of battery Internal resistance R₀, polarization resistance Rp and the polarization capacity Cp of battery.

The parameter identification of the electrokinetic cell model of I-2 first structure changes filtering

Use change the first structure filtering that the discrete model of battery equivalent model is carried out parameter identification.Select according to formula (3) Select systematic state variable X_k=[a_k,b_k,c_k,(1-a)U_oc,k]^T, obtaining state equation is:

$\left\{\begin{array}{c}{X}_k=X_{k-1}+w_{k-1}\\ Z_k=U_k={\mathrm{Cx}}_k+v_k\end{array}\right.---\left(4\right)$

Wherein k is current time, X_k∈R_n× 1 is system mode vector；Z_k∈R_m× 1 is measuring state variable；w_kAnd v_k It is respectively system zero average stochastic process noise and measures noise.C is to measure equation coefficient, C=[U_k-1,I_k,I_k-1,1]。

Parameter identification process under structure changes filtering-I algorithm is as follows:

$\left[\begin{array}{c}{\hat{z}}_{k|k-1}\\ {\hat{z}}_{k|k}\end{array}\right]=C\left[\begin{array}{c}{\hat{X}}_{k|k-1}\\ {\hat{X}}_{k|k}\end{array}\right]---\left(5\right)$

${\hat{X}}_{k|k-1}={\hat{X}}_{k-1|k-1}---\left(6\right)$

${\hat{X}}_{k|k}={\hat{X}}_{k|k-1}+K_k---\left(7\right)$

First structure changes filteringIt is the systematic state variable X correction value in the k-1 moment,It it is system mode Variable X in the correction value in k moment,It it is the systematic state variable X predictive value in the k moment；It is measuring state variable Z Predictive value in the k moment；It it is the measuring state variable z correction value in the k moment；K_kIt is that the parameter identification correction in the k moment increases Amount, i.e. systematic state variable X is at the predictive value in k momentCorrection.

First structure changes filtering is by adjusting the correction increment K of structure changes filtering_k, constantly revise state variable X in the k moment Predictive valueThe correction increment K in k moment_kFor:

$\left\{\begin{array}{c}{e}_{z_{k|k-1}}=z_k-{\hat{z}}_{k|k-1}\\ e_{z_{k-1|k-1}}=z_{k-1}-{\hat{z}}_{k-1|k-1}\end{array}\right.---\left(9\right)$

(8), z in (9) formula_kThe measured value in terminal voltage k moment when being battery operation, i.e. U_k；It it is the current k moment Error between actual value and the k moment predictive value of measuring state variable z；It is the true of k-1 moment measuring state variable Error between real-valued and k-1 moment predictive value；C^-1It it is the inverse matrix of vector C；β, γ are constant values, span be 0～ 1, and the value of β directly affects the accuracy of structure changes filtering parameter identification；ο is Schur product, i.e. two matrix corresponding elements The result being multiplied；Sat is saturation function, and wherein Ψ is the smooth boundary layer thickness of the first structure changes filtering, the first structure changes filter The vector definition of the sat saturation function of ripple is as follows:

$sat(e_{z_{k|k-1}},\mathrm{\&psi;})={\left[\begin{array}{ccc}sat(e_{z_{1,k|k-1}},{\mathrm{\&psi;}}_1)& \mathrm{...}& sat(e_{z_{n,k|k-1}},{\mathrm{\&psi;}}_n)\end{array}\right]}^T---\left(10\right)$

Wherein the saturation function sat of the first structure changes filtering is defined as follows:

$sat(e_{z_{i,k|k-1}},{\mathrm{\&psi;}}_i)=\left\{\begin{array}{cc}{e}_{z_{i,k|k-1}}/{\mathrm{\&psi;}}_i& e_{z_{i,k|k-1}}\&le;{\mathrm{\&psi;}}_i\\ sign\left(e_{z_{i,k|k-1}}\right)& e_{z_{i,k|k-1}}>{\mathrm{\&psi;}}_i\end{array}\right.---\left(11\right)$

Wherein Ψ_iIt is for deviationThe border be given to introduce boundary region, | Ψ_i| for the thickness of boundary region, It is taken as constant, | Ψ_i|=0.01～0.03；Sign represents sign function:

$sign\left(e_{z_{i,k|k-1}}\right)=\left\{\begin{array}{c}{e}_{z_{i,k|k-1}}>0\\ e_{z_{i,k|k-1}}=0\\ e_{z_{i,k|k-1}}<0\end{array}\right..$

Filtered by above first structure changes, obtain system mode vector X_k=[a_k,b_k,c_k,(1-a)U_oc,k]^TEstimation Value isThe parameter value of electrokinetic cell model is obtained: ohmic internal resistance R from system mode vector estimated value₀, polarization resistance Rp, pole Change electric capacity Cp and estimated value U of k moment open-circuit voltage_{Oc, k}, i.e. OCV.

II, the matching of electrokinetic cell open-circuit voltage OCV-SOC relation

The high price approximation by polynomi-als fit mathematics relationship model of OCV-SOC is as follows:

$\begin{array}{c}{U}_{oc,k}=g\left({\mathrm{SOC}}_k\right)\\ =h_1{{\mathrm{SOC}}_k}^8+h_2{{\mathrm{SOC}}_k}^7+h_3{{\mathrm{SOC}}_k}^6+h_4{{\mathrm{SOC}}_k}^5+h_5{{\mathrm{SOC}}_k}^4\\ +h_6{{\mathrm{SOC}}_k}^3+h_7{{\mathrm{SOC}}_k}^2+h_8{{\mathrm{SOC}}_k}^1+h_9\end{array}---\left(12\right)$

In formula: h₁～h₉For the coefficient under OCV-SOC high price fitting of a polynomial, obtain after approaching matching: h₁=2.10 × 10³, h₂=-7.38 × 10³, h₃=9.98 × 10³, h₄=-6.23 × 10³, h₅=1.40 × 10³, h₆=3.26 × 10², h₇=-2.40 ×10², h₈=47.98, h₉=22.27.SOC_kRepresent under DST operating mode, use high precision electro flow measurement, fixed by the SOC generally acknowledged Justice method obtain in k moment battery dump energy value.

III, the SOC estimation method of the second structure changes filtering

Select polarization capacity C in SOC and battery model_pTerminal voltage U_p,kSystem mode as the second structure changes filtering Variable, i.e. XX_k=[SOC_k U_p,k]^T, the state equation of system and measurement equation are as follows:

$\left\{\begin{array}{c}{\mathrm{XX}}_k={\mathrm{AXX}}_{k-1}+{\mathrm{BI}}_{k-1}+{\mathrm{ww}}_k\\ {\mathrm{ZZ}}_k=U_k=U_{oc,k}-R_0{I}_k-U_{P,k}+{\mathrm{vv}}_k\end{array}\right.---\left(13\right)$

Wherein:

(13) in formula: T is the sampling period；Q_NFor battery rated capacity；η is discharge and recharge coulombic efficiency；R_pRepresent the pole of battery Change internal resistance；C_pRepresent the polarization capacity of battery；U_oc,kRepresent the open-circuit voltage of k moment battery；R_oRepresent the ohmic internal resistance of battery； I_kRepresent that the k moment flows through ohmic internal resistance R_oElectric current；U_kRepresent k moment terminal voltage during battery operation；ZZ_kIt it is the second structure changes The measuring state variable of filtering, ww_kAnd vv_kIt is respectively system zero average stochastic process noise and the measurement of the second structure changes filtering Noise, the system zero average stochastic process noise w that its variance also filters with the first structure changes_kWith measurement noise v_kVariance different, Mainly caused by sensor accuracy, model error etc..

Formula (12) is substituted into measurement equation (13) arrange:

$\begin{array}{c}{U}_k={\mathrm{CCX}}_k+h_1{{\mathrm{SOC}}_k}^8+h_2{{\mathrm{SOC}}_k}^7+h_3{{\mathrm{SOC}}_k}^6+h_4{{\mathrm{SOC}}_k}^5\\ +h_5{{\mathrm{SOC}}_k}^4+h_6{{\mathrm{SOC}}_k}^3+h_7{{\mathrm{SOC}}_k}^2+h_9-R_0{I}_k\end{array}---\left(14\right)$

Wherein CC is the coefficient measuring equation, CC=[h₈-1]；SOC_kRepresent the dump energy estimated value of k moment battery.

Obtain according to the first structure changes Filtering Formula (7) in I-2 step:

When state updatesIn KK_kFor revising increment, i.e.Correction value:

Wherein:It is the systematic state variable X X correction value in the k moment of the second structure changes filtering,It is The system state variables xx of two structure changes filtering is at the predictive value in k moment；CC^-1It it is the converse matrix of vector CC；

$\left\{\begin{array}{c}{e}_{{\mathrm{zz}}_{k/k-1}}=Z{Z}_k-Z{\hat{Z}}_{k/k-1}\\ e_{{\mathrm{zz}}_{k-1/k-1}}={\mathrm{ZZ}}_{k-1}-Z{\hat{Z}}_{k-1/k-1}\end{array}\right.$

It it is the mistake k moment that the second structure changes filters measuring between actual value and the predictive value of system state variables Difference；Be k-1 constantly of the second structure changes filtering measures the actual value of system state variables and revised predictive value it Between error；ο is Schur product；Sat is saturation function, and wherein ψ ψ is the smooth boundary layer thickness the of the second structure changes filtering The vector of the sat saturation function of two structure changes filtering is defined as follows:

$sat(e_{{\mathrm{zz}}_{k|k-1}},\mathrm{\&psi;}\mathrm{\&psi;})={\left[\begin{array}{ccc}sat(e_{{\mathrm{zz}}_{1,k|k-1}},{\mathrm{\&psi;\&psi;}}_1)& \mathrm{...}& sat(e_{{\mathrm{zz}}_{n,k|k-1}},{\mathrm{\&psi;\&psi;}}_n)\end{array}\right]}^T,$

Wherein the saturation function sat of the second structure changes filtering is defined as follows:

$sat(e_{{\mathrm{zz}}_{i,k|k-1}},{\mathrm{\&psi;\&psi;}}_i)=\left\{\begin{array}{cc}{e}_{{\mathrm{zz}}_{i,k|k-1}}/{\mathrm{\&psi;\&psi;}}_i& e_{{\mathrm{zz}}_{i,k|k-1}}\&le;{\mathrm{\&psi;\&psi;}}_i\\ sign\left(e_{{\mathrm{zz}}_{i,k|k-1}}\right)& e_{{\mathrm{zz}}_{i,k|k-1}}>{\mathrm{\&psi;\&psi;}}_i\end{array}\right.,$

Wherein ψ ψ_iIt is for deviationThe border introducing boundary region and provide, | ψ ψ_i| for the thickness of boundary region, this example It is 0.02；Sign represents sign function, and its rule is as follows:

$sign\left(e_{{\mathrm{zz}}_{i,k|k-1}}\right)=\left\{\begin{array}{c}{e}_{{\mathrm{zz}}_{i,k|k-1}}>0\\ e_{{\mathrm{zz}}_{i,k|k-1}}=0\\ e_{{\mathrm{zz}}_{i,k|k-1}}<0\end{array}\right..$

(15) in the correction incremental computations formula of formula, β β be second structure changes filtering terminal voltage estimated value and terminal voltage true Error between measured valueCoefficient, γ γ be second structure changes filtering terminal voltage correction value and terminal voltage truly survey Error between valueCoefficient.

Parameter beta β value is mainly adjusted by this example.

The fuzzy rule of β β value is as follows:

$\mathrm{\&beta;}\mathrm{\&beta;}=\left\{\begin{array}{cc}1& e_{{\mathrm{zz}}_{k/k-1}}\&GreaterEqual;0.034U_N\\ 0.1\&CenterDot;e_{{\mathrm{zz}}_{k/k-1}}& 0.034U_N>e_{{\mathrm{zz}}_{k/k-1}}>-0.034U_N\\ 1& e_{{\mathrm{zz}}_{k/k-1}}\&le;-0.034U_N\end{array}\right.---\left(16\right)$

In formulaIt it is the error between actual value and the predictive value measuring system state variables；U_NSpecified electricity for battery Pressure, this example is 24V.

Adaptive fuzzy structure changes filtering being carried out parameter by formula (16) adjusts, and is fuzzy variable structure filtering.Should Fuzzy variable structure filtering and the structure changes for battery parameter identification filter to combine and obtain the SOC estimation of electrokinetic cell Value, as shown in Figure 2.

The estimating system embodiment of the power battery charged state SOC of double structure changes filtering

The estimating system embodiment of the power battery charged state SOC of this pair of structure changes filtering, including microprocessor, modulus Modular converter, current sensor, voltage sensor.This example electrokinetic cell is 24V, 22Ah.

Voltage sensor and current sensor are installed on electrokinetic cell port, respectively detection electrokinetic cell terminal voltage and port Electric current.Voltage, current sensor connect the general-purpose interface of microprocessor through analog-to-digital conversion module, and output end of microprocessor connects Display screen, the estimated value of the battery charge state that Real time displaying is current, the i.e. current SOC value estimated.Output end of microprocessor is also Connect CAN interface, be connected with the CAN controller of automobile, be i.e. connected with other electrical communications of automobile.

Microprocessor contains data storage and program storage, and described program storage is contained within structure changes filtering power Battery parameter identification module, open-circuit voltage and SOC relation fitting module and parameter adaptive fuzzy adjustment module.Data store The data that in the device storage voltage of electrokinetic cell, current detecting data and program storage, each module produces during calculating.

Voltage sensor and current sensor detection obtain the simulation letter of the electric current of current power battery terminal voltage and port Number through analog-to-digital conversion module be converted to correspondence digital signal, send into microprocessor, microprocessor call structure changes filtering power Battery parameter identification module, open-circuit voltage and SOC relation fitting module and parameter adaptive fuzzy adjustment module, according to currently Terminal voltage and current value, estimate by the method for estimation embodiment of the power battery charged state of above-mentioned pair of structure changes filtering Calculate, obtain the estimated value of current battery charge state.

Above-described embodiment, only further describe the purpose of the present invention, technical scheme and beneficial effect is concrete Individual example, the present invention is not limited to this.All made within the scope of disclosure of the invention any amendment, equivalent, change Enter, within being all contained in protection scope of the present invention.

Claims (7)

1. a method of estimation of the power battery charged state SOC of double structure changes filtering, key step is as follows:

I, the parameter identification of electrokinetic cell

I-1, electrokinetic cell discrete model

Using equivalent model is battery equivalent model, describes static state and the dynamic property of battery；The polarization resistance R of battery_pWith battery Polarization capacity C_pComposition single order reinforced concrete structure in parallel, represents the polarization reaction of battery, and RC both end voltage is U_pT () represents battery-end Voltage；Concatenation ohmage R₀, flow through ohmic internal resistance R₀Electric current be i (t), Uoc (t) be the open-circuit voltage OCV of battery, sampling Obtain battery terminal voltage U (t) and flow through ohmic internal resistance R₀Electric current i (t)；

Described battery equivalent model mathematic(al) representation is as follows:

$\left\{\begin{array}{c}\frac{{\mathrm{dU}}_p\left(t\right)}{dt}=-\frac{U_p\left(t\right)}{R_p{C}_p}+\frac{i\left(t\right)}{C_p}\\ U\left(t\right)=U_{OC}\left(t\right)-R_{0}i\left(t\right)-U_p\left(t\right)\end{array}\right.---\left(1\right)$

Use the backward difference alternative approach battery model discretization to formula (1):

U_k-U_oc,k=a (U_k-1-U_oc,k-1)+bI_k+cI_k-1 (2)

Wherein k is current time, and k-1 is previous moment, and a, b and c are respectively model parameter:

Formula (2) is arranged the discrete model of electrokinetic cell equivalent model is:

U_k=a_kU_k-1+b_kI_k+c_kI_k-1+(1-a_k)U_oc,k (3)

Wherein, a_k、b_k、c_kAs follows with battery backward difference model parameter relation:

$R_0=\frac{c_k}{a_k},$

$R_p=\frac{-a_k{b}_k-c_k}{a_k(1-a_k)},$

$C_p=\frac{{{\mathrm{Ta}}_k}^2}{-a_k{b}_k-c_k},$

In formula, T is the sampling period, and T is 0.5 second to 2 seconds；

By the parameter identification of electrokinetic cell discrete model, obtain electrokinetic cell parameter open-circuit voltage Uoc, the ohmic internal resistance of battery R₀, polarization resistance Rp and the polarization capacity Cp of battery；

The parameter identification of the electrokinetic cell model of I-2 first structure changes filtering

Use change the first structure filtering that the discrete model of battery equivalent model is carried out parameter identification；System is selected according to formula (3) System state variable X_k=[a_k,b_k,c_k,(1-a)U_oc,k]^T, obtaining state equation is:

$\left\{\begin{array}{c}{X}_k=X_{k-1}+w_{k-1}\\ Z_k=U_k={\mathrm{Cx}}_k+v_k\end{array}\right.---\left(4\right)$

Wherein k is current time, X_k∈R_n× 1 is system mode vector；Z_k∈R_m× 1 is measuring state variable；w_kAnd v_kRespectively For system zero average stochastic process noise and measurement noise；C is to measure equation coefficient, C=[U_k-1,I_k,I_k-1,1]；

Parameter identification process under structure changes filtering-I algorithm is as follows:

$\left[\begin{array}{c}{\hat{z}}_{k|k-1}\\ {\hat{z}}_{k|k}\end{array}\right]=C\left[\begin{array}{c}{\hat{X}}_{k|k-1}\\ {\hat{X}}_{k|k}\end{array}\right]---\left(5\right)$

${\hat{X}}_{k|k-1}={\hat{X}}_{k-1|k-1}---\left(6\right)$

${\hat{X}}_{k|k}={\hat{X}}_{k|k-1}+K_k---\left(7\right)$

First structure changes filteringIt is the systematic state variable X correction value in the k-1 moment,It it is system state variables X in the correction value in k moment,It it is the systematic state variable X predictive value in the k moment；It is that measuring state variable z is at k The predictive value in moment；It it is the measuring state variable z correction value in the k moment；K_kIt is that the parameter identification correction in the k moment increases Amount, i.e. systematic state variable X is at the predictive value in k momentCorrection；

First structure changes filtering is by adjusting the correction increment K of structure changes filtering_k, continuous correction state variable X is pre-the k moment Measured valueThe correction increment K in k moment_kFor:

$\left\{\begin{array}{c}{e}_{z_{k|k-1}}=z_k-{\hat{z}}_{k|k-1}\\ e_{z_{k-1|k-1}}=z_{k-1}-{\hat{z}}_{k-1|k-1}\end{array}\right.---\left(9\right)$

(8), z in (9) formula_kThe measured value in terminal voltage k moment when being battery operation, i.e. U_k；It is the current k moment to measure shape Error between actual value and the k moment predictive value of state variable z；Be k-1 moment measuring state variable actual value with Error between k-1 moment predictive value；C^-1It it is the inverse matrix of vector C；β, γ are constant values, and span is 0～1, and β Value directly affects the accuracy of structure changes filtering parameter identification；ο is Schur product, and i.e. two matrix corresponding elements are multiplied Result；Sat is saturation function, and wherein Ψ is the smooth boundary layer thickness of the first structure changes filtering, the sat of the first structure changes filtering The vector definition of saturation function is as follows:

$sat(e_{z_{k|k-1}},\mathrm{\&psi;})={\left[\begin{array}{ccc}sat(e_{z_{1,k|k-1}},{\mathrm{\&psi;}}_1)& ...& sat(e_{z_{n,k|k-1}},{\mathrm{\&psi;}}_n)\end{array}\right]}^T---\left(10\right)$

Wherein the saturation function sat of the first structure changes filtering is defined as follows:

$sat(e_{z_i{,}_{k|k-1}},{\mathrm{\&psi;}}_i)=\left\{\begin{array}{cc}{e}_{z_{i,k|k-1}}/{\mathrm{\&psi;}}_i& e_{z_{i,k|k-1}}\&le;{\mathrm{\&psi;}}_i\\ sign\left(e_{z_{i,k|k-1}}\right)& e_{z_{i,k|k-1}}>{\mathrm{\&psi;}}_i\end{array}\right.---\left(11\right)$

Wherein Ψ_iIt is for deviationThe border be given to introduce boundary region, | Ψ_i| for the thickness of boundary region, often it is taken as Amount, | Ψ_i|=0.01～0.03；Sign represents sign function:

$sign\left(e_{z_{i,k|k-1}}\right)=\left\{\begin{array}{c}{e}_{z_{i,k|k-1}}>0\\ e_{z_{i,k|k-1}}=0\\ e_{z_{i,k|k-1}}<0\end{array}\right.;$

Filtered by above first structure changes, obtain system mode vector X_k=[a_k,b_k,c_k,(1-a)U_oc,k]^TEstimated value beThe parameter value of electrokinetic cell model is obtained: ohmic internal resistance R from system mode vector estimated value₀, polarization resistance Rp, polarization electricity Hold estimated value U of Cp and k moment open-circuit voltage_{Oc, k}, i.e. OCV；

II, the matching of electrokinetic cell open-circuit voltage OCV-SOC relation

The high price approximation by polynomi-als fit mathematics relationship model of OCV-SOC is as follows:

$\begin{array}{c}{U}_{oc,k}=g\left({\mathrm{SOC}}_k\right)\\ =h_1{{\mathrm{SOC}}_k}^8+h_2{{\mathrm{SOC}}_k}^7+h_3{{\mathrm{SOC}}_k}^6+h_4{{\mathrm{SOC}}_k}^5+h_5{{\mathrm{SOC}}_k}^4\\ +h_6{{\mathrm{SOC}}_k}^3+h_7{{\mathrm{SOC}}_k}^2+h_8{{\mathrm{SOC}}_k}^1+h_9\end{array}---\left(12\right)$

In formula: h₁～h₉For the coefficient under OCV-SOC high price fitting of a polynomial, obtain after approaching matching: h₁=2.10 × 10³, h₂=- 7.38×10³, h₃=9.98 × 10³, h₄=-6.23 × 10³, h₅=1.40 × 10³, h₆=3.26 × 10², h₇=-2.40 × 10², h₈=47.98, h₉=22.27；SOC_kRepresent in k moment battery dump energy value；

III, the SOC estimation method of the second structure changes filtering

Select polarization capacity C in SOC and battery model_pTerminal voltage U_p,kAs the system state variables of the second structure changes filtering, I.e. XX_k=[SOC_k U_p,k]^T, the state equation of system and measurement equation are as follows:

$\left\{\begin{array}{c}{\mathrm{XX}}_k={\mathrm{AXX}}_{k-1}+{\mathrm{BI}}_{k-1}+{\mathrm{ww}}_k\\ {\mathrm{ZZ}}_k=U_k=U_{oc,k}-R_0{I}_k-U_{P,k}+{\mathrm{vv}}_k\end{array}\right.---\left(13\right)$

Wherein:

(13) in formula: T is the sampling period；Q_NFor battery rated capacity；η is discharge and recharge coulombic efficiency；R_pIn representing the polarization of battery Resistance；C_pRepresent the polarization capacity of battery；U_oc,kRepresent the open-circuit voltage of k moment battery；R_oRepresent the ohmic internal resistance of battery；I_kTable Show that the k moment flows through ohmic internal resistance R_oElectric current；U_kRepresent k moment terminal voltage during battery operation；zz_kIt is the second structure changes filtering Measuring state variable, ww_kAnd vv_kIt is respectively the system zero average stochastic process noise of the second structure changes filtering and measures noise；

Formula (12) is substituted into measurement equation (13) arrange:

$\begin{array}{c}{U}_k={\mathrm{CCX}}_k+h_1{{\mathrm{SOC}}_k}^8+h_2{{\mathrm{SOC}}_k}^7+h_3{{\mathrm{SOC}}_k}^6+h_4{{\mathrm{SOC}}_k}^5\\ +h_5{{\mathrm{SOC}}_k}^4+h_6{{\mathrm{SOC}}_k}^3+h_7{{\mathrm{SOC}}_k}^2+h_9-R_0{I}_k\end{array}---\left(14\right)$

Wherein CC is the coefficient measuring equation, CC=[h₈-1]；SOC_kRepresent the dump energy estimated value of k moment battery；

Obtain according to the first structure changes Filtering Formula (7) in I-2 step:

When state updatesIn KK_kFor revising increment, i.e.Correction value:

Wherein:It is the systematic state variable X X correction value in the k moment of the second structure changes filtering,It it is the second change The systematic state variable X X of structure filtering is at the predictive value in k moment；CC^-1It it is the converse matrix of vector CC；

$\left\{\begin{array}{c}{e}_{{\mathrm{zz}}_{k/k-1}}={\mathrm{ZZ}}_k-Z{\hat{Z}}_{k/k-1}\\ e_{{\mathrm{zz}}_{k-1/k-1}}={\mathrm{ZZ}}_{k-1}-Z{\hat{Z}}_{k-1/k-1}\end{array}\right.$

It it is the error k moment that the second structure changes filters measuring between actual value and the predictive value of system state variables；It is to measure between the actual value of system state variables and revised predictive value in the k-1 moment that the second structure changes filters Error；ο is Schur product；Sat is saturation function, and wherein ψ ψ is that the smooth boundary layer thickness second that the second structure changes filters becomes The vector of the sat saturation function of structure filtering is defined as follows:

$sat(e_{{\mathrm{zz}}_{k|k-1}},\mathrm{\&psi;}\mathrm{\&psi;})={\left[\begin{array}{ccc}sat(e_{{\mathrm{zz}}_{1,k|k-1}},{\mathrm{\&psi;\&psi;}}_1)& ...& sat(e_{{\mathrm{zz}}_{n,k|k-1}},{\mathrm{\&psi;\&psi;}}_n)\end{array}\right]}^T,$

Wherein the saturation function sat of the second structure changes filtering is defined as follows:

$sat(e_{{\mathrm{zz}}_i{,}_{k|k-1}},{\mathrm{\&psi;\&psi;}}_i)=\left\{\begin{array}{cc}{e}_{{\mathrm{zz}}_{i,k|k-1}}/{\mathrm{\&psi;\&psi;}}_i& e_{{\mathrm{zz}}_{i,k|k-1}}\&le;{\mathrm{\&psi;\&psi;}}_i\\ sign\left(e_{{\mathrm{zz}}_{i,k|k-1}}\right)& e_{{\mathrm{zz}}_{i,k|k-1}}>{\mathrm{\&psi;\&psi;}}_i\end{array}\right.,$

Wherein ψ ψ_iIt is for deviationThe border introducing boundary region and provide, | ψ ψ_i| for the thickness of boundary region, it is taken as constant； Sign represents sign function, and its rule is as follows:

$sign\left(e_{{\mathrm{zz}}_{i,k|k-1}}\right)=\left\{\begin{array}{c}{e}_{{\mathrm{zz}}_{i,k|k-1}}>0\\ e_{{\mathrm{zz}}_{i,k|k-1}}=0\\ e_{{\mathrm{zz}}_{i,k|k-1}}<0\end{array}\right.$

(15) in the correction incremental computations formula of formula, β β be second structure changes filtering terminal voltage estimated value and terminal voltage truly measure Error between valueCoefficient, γ γ be second structure changes filtering terminal voltage correction value and terminal voltage true measurement Between errorCoefficient.

The method of estimation of the power battery charged state SOC of the most according to claim 1 pair of structure changes filtering, its feature exists In:

Described sampling period T is 0.5 second to 2 seconds.

The method of estimation of the power battery charged state SOC of the most according to claim 2 pair of structure changes filtering, its feature exists In:

The fuzzy rule of the β β value in described step III is as follows:

$\mathrm{\&beta;}\mathrm{\&beta;}=\left\{\begin{array}{cc}1& e_{{\mathrm{zz}}_{k/k-1}}\&GreaterEqual;0.034U_N\\ 0.1\&CenterDot;e_{{\mathrm{zz}}_{k/k-1}}& 0.034U_N>e_{{\mathrm{zz}}_{k/k-1}}>-0.034U_N\\ 1& e_{{\mathrm{zz}}_{k/k-1}}\&le;-0.034U_N\end{array}\right.---\left(16\right)$

In formulaIt it is the error between actual value and the predictive value measuring system state variables；U_NRated voltage for battery.

The estimation side of the power battery charged state SOC of the most according to any one of claim 1 to 3 pair of structure changes filtering Method, it is characterised in that:

In described step III | ψ ψ_i| value 0.01～0.03.

The one of the method for estimation design of the power battery charged state SOC of the most according to claim 3 pair of structure changes filtering Plant the estimating system of the power battery charged state of double structure changes filtering, including microprocessor, analog-to-digital conversion module, current sense Device, voltage sensor；Voltage sensor and current sensor are respectively arranged in electrokinetic cell port to be detected, detect power current Pond terminal voltage and the electric current of port, voltage, current sensor connect microprocessor through analog-to-digital conversion module, and described microprocessor is defeated Go out the estimated value of present battery state-of-charge；It is characterized in that:

Described microprocessor contains data storage and program storage, and described program storage is contained within structure changes filtering power Battery parameter identification module, open-circuit voltage and SOC relation fitting module and parameter adaptive fuzzy adjustment module；Described data The number that in the memorizer storage voltage of electrokinetic cell, current detecting data and program storage, each module produces during calculating According to.

The estimating system of the power battery charged state SOC of the most according to claim 5 pair of structure changes filtering, its feature exists In:

Described microprocessor is connected with display screen, the estimated value of the battery charge state that Real time displaying is current.

The estimating system of the power battery charged state SOC of the most according to claim 5 pair of structure changes filtering, its feature exists In:

Described microprocessor is furnished with CAN interface.