CN105891727A

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

Info

Publication number: CN105891727A
Application number: CN201610412667.1A
Authority: CN
Inventors: 党选举; 姜辉; 伍锡如; 李爽; 张向文; 李珊; 朱国魂; 叶懋; 莫太平; 王金辉; 王涵正
Original assignee: Guilin University of Electronic Technology
Current assignee: Guilin University of Electronic Technology
Priority date: 2016-06-13
Filing date: 2016-06-13
Publication date: 2016-08-24
Anticipated expiration: 2036-06-13
Also published as: CN105891727B

Abstract

The present invention is a method and system for estimating SOC of power battery state of charge with double variable structure filtering. The relationship between the open-circuit voltage OCV and SOC of the power battery, and then use the second variable structure filter to estimate the SOC. The value of variable structure filter parameter ββ has a great influence on the correction increment, and fuzzy rules are introduced to adjust ββ adaptively. The voltage and current sensors of the estimation system are installed on the power battery to be tested, and the microprocessor contains a calculation module for executing the method, and the microprocessor displays the current estimated SOC value online, and can be connected with the CAN controller of the vehicle. The present invention identifies power battery parameters on-line; self-adaptive fuzzy adjustment of correction increment of variable structure filtering parameters has simple estimation method, small calculation amount, easy realization, high precision and little dependence on SOC initial value.

Description

A method and system for estimating the state of charge of a power battery with double variable structure filtering

技术领域technical field

本发明涉及电动汽车的动力电池状态估计领域，具体为一种双变结构滤波的动力电池荷电状态的估计方法与系统。The invention relates to the field of power battery state estimation of electric vehicles, in particular to a method and system for estimating the state of charge of a power battery with double variable structure filtering.

背景技术Background technique

新能源汽车是解决传统汽车对环境污染的一种有效方法，动力电池是新能源汽车核心关键技术之一。高精度动力电池的荷电状态(State of Charge，简称SOC)的估算值是动力汽车行驶过程中的重要参数，也是电动汽车行驶状态的重要依据。New energy vehicles are an effective way to solve the environmental pollution caused by traditional vehicles, and power batteries are one of the core technologies of new energy vehicles. The estimated value of the state of charge (SOC) of the high-precision power battery is an important parameter in the driving process of the power vehicle, and it is also an important basis for the driving state of the electric vehicle.

现常用动力电池荷电状态估算方法有：安时积分法、开路电压法、卡尔曼滤波法和粒子滤波法等。实际使用时这些方法均有难以克服的缺陷。安时积分法受车载传感器精度的限制，容易导致累计误差，造成动力电池的荷电状态SOC估算误差很大。开路电压法的开路电压数据的采集需要电池长时间的静置，才能准确测量，因而不适合实际行车过程中的SOC在线估计。卡尔曼滤波法多用于线性的动力电池的荷电状态SOC估计。扩展卡尔曼滤波法、无迹卡尔曼滤波法等将动力电池的荷电状态SOC作为一个状态变量建立状态空间模型，通过递推得到动力电池的荷电状态SOC最小方差估计值，但这些改进的卡尔曼滤波法对模型的依赖性很高；粒子滤波法需要用大量的样本数量才能有效地近似系统的后验概率密度，运算量相当大。The commonly used methods for estimating the state of charge of power batteries include: ampere-hour integral method, open circuit voltage method, Kalman filter method and particle filter method. These methods have insurmountable defects in actual use. The ampere-hour integration method is limited by the accuracy of on-board sensors, which can easily lead to cumulative errors, resulting in a large error in the estimation of the state of charge SOC of the power battery. The open-circuit voltage data acquisition of the open-circuit voltage method requires the battery to stand still for a long time before it can be accurately measured, so it is not suitable for online SOC estimation in the actual driving process. The Kalman filter method is mostly used for SOC estimation of the linear power battery state of charge. The extended Kalman filter method, the unscented Kalman filter method, etc. use the state of charge SOC of the power battery as a state variable to establish a state space model, and obtain the minimum variance estimate of the state of charge SOC of the power battery through recursion, but these improvements The Kalman filter method is highly dependent on the model; the particle filter method requires a large number of samples to effectively approximate the posterior probability density of the system, and the amount of calculation is quite large.

总之现有的动力电池的荷电状态SOC估算方法由于误差累计、动力电池的非线性等影响估算精度，目前尚未有SOC在线估算精度高、收敛速度较快，且对初值精度依赖度低的动力电池的荷电状态SOC估计方法。In short, the existing SOC estimation method of the power battery is due to the influence of error accumulation and nonlinearity of the power battery on the estimation accuracy. At present, there is no online SOC estimation method with high accuracy, fast convergence speed, and low dependence on the accuracy of the initial value. SOC estimation method for power battery state of charge.

发明内容Contents of the invention

本发明的目的是设计一种采用双变结构滤波的动力电池荷电状态的估计方法，采用一个变结构滤波对电池模型参数辨识，采用另一模糊变结构滤波，对电池模型参数的修正增量进行自适应模糊调整，本方法不仅保证SOC估算值收敛的有效，而且保证估算值的精度。双变结构滤波结构简洁，能适应SOC不同初值状态，实现SOC的高精度估计。The purpose of the present invention is to design a method for estimating the state of charge of a power battery using a double variable structure filter. One variable structure filter is used to identify the battery model parameters, and another fuzzy variable structure filter is used to correct the increment of the battery model parameters. By performing adaptive fuzzy adjustment, this method not only ensures the validity of the convergence of the estimated value of SOC, but also ensures the accuracy of the estimated value. The dual-variable filter has a simple structure and can adapt to different initial states of the SOC to achieve high-precision estimation of the SOC.

本发明的另一目的是根据一种双变结构滤波的动力电池荷电状态的估计方法设计双变结构滤波的动力电池荷电状态的估计系统，采用变结构滤波对电池模型参数辨识，经具有修正增量的自适应调整的模糊变结构滤波实现SOC的估计。Another object of the present invention is to design a dual variable structure filter power battery state of charge estimation system based on a method for estimating the state of charge of a power battery with a double variable structure filter, and use a variable structure filter to identify battery model parameters. The estimation of SOC is realized by the fuzzy variable structure filtering with self-adaptive adjustment of correction increment.

本发明设计的一种双变结构滤波的动力电池荷电状态SOC的估计方法主要步骤如下：The main steps of a method for estimating the state of charge SOC of a power battery with double variable structure filtering designed by the present invention are as follows:

Ⅰ、动力电池的参数辨识Ⅰ. Parameter identification of power battery

Ⅰ-1、动力电池离散模型Ⅰ-1. Discrete model of power battery

本发明采用目前最广泛使用的Thevenin模型为电池等效模型，描述电池的静态和动态性能。电池的极化内阻R_p与电池的极化电容C_p并联构成一阶RC结构，表示电池的极化反应，RC两端电压为U_p(t)表示电池端电压；串接欧姆电阻R₀，流过欧姆内阻R₀的电流为i(t)，Uoc(t)为电池的开路电压OCV(Open circuit voltage)，采样得到电池端电压U(t)和流过欧姆内阻R₀的电流i(t)。The present invention adopts the most widely used Thevenin model as the battery equivalent model to describe the static and dynamic performance of the battery. The polarized internal resistance R _p of the battery and the polarized capacitance C _p of the battery are connected in parallel to form a first-order RC structure, which represents the polarization reaction of the battery. The voltage across the RC is U _p (t) represents the terminal voltage of the battery; ₀ , the current flowing through the ohmic internal resistance R ₀ is i(t), Uoc(t) is the open circuit voltage OCV (Open circuit voltage) of the battery, and the battery terminal voltage U(t) and the current flowing through the ohmic internal resistance R ₀ are obtained by sampling The current i(t).

所述电池有效模型数学表达式如下：The mathematical expression of the battery effective model is as follows:

$\{\begin{matrix} \frac{{dU U}_{p p} ((11))}{d d t t} = = - - \frac{{U u}_{p p} ((t t))}{{R R}_{p p} {C C}_{p p}} + + \frac{i i ((t t))}{{C C}_{p p}} \\ U u ((t t)) = = {U u}_{O o C C} ((t t)) - - {R R}_{00} i i ((11)) - - {U u}_{p p} ((t t)) \end{matrix} - - - - - - ((11))$

采用后向差分变换方法对公式(1)的电池模型离散化，得：The battery model of formula (1) is discretized by using the backward differential transformation method, and it is obtained:

U_k-U_oc,k＝a(U_k-1-U_oc,k-1)+bI_k+cI_k-1 (2)U _k -U _oc,k ＝a(U _k-1 -U _oc,k-1 )+bI _k +cI _k-1 (2)

其中k是当前时刻，k-1是前一时刻，a、b和c分别为模型参数。Where k is the current moment, k-1 is the previous moment, and a, b, and c are model parameters respectively.

考虑到动力电池的充放电过程缓慢，开路电压Uoc变化较小，即U_oc,k＝U_oc,k-1，以此，对公式(2)整理得动力电池等效模型的离散模型为：Considering that the charging and discharging process of the power battery is slow, the change of the open circuit voltage Uoc is small, that is, U _oc,k = U _oc,k-1 , based on this, the discrete model of the equivalent model of the power battery according to formula (2) is:

U_k＝a_kU_k-1+b_kI_k+c_kI_k-1+(1-a_k)U_oc,k (3)U _k ＝a _k U _k-1 +b _k I _k +c _k I _k-1 +(1-a _k )U _oc,k (3)

其中，a_k、b_k、c_k与电池后向差分模型参数关系如下：Among them, the relationship between a _k , b _k , c _k and the parameters of the battery backward differential model is as follows:

${R R}_{00} = = \frac{{c c}_{k k}}{{a a}_{k k}},,$

${R R}_{p p} = = \frac{- - {a a}_{k k} {b b}_{k k} - - {c c}_{k k}}{{a a}_{k k} ((11 - - {a a}_{k k}))},,$

${C C}_{p p} = = \frac{{Ta Ta}_{k k}^{22}}{- - {a a}_{k k} {b b}_{k k} - - {c c}_{k k}},,$

式中T为采样周期，T为0.5秒至2秒。In the formula, T is the sampling period, and T is 0.5 seconds to 2 seconds.

通过动力电池离散模型的参数辨识，得到动力电池参数开路电压Uoc、电池的欧姆内阻R₀、极化内阻Rp及电池的极化电容Cp。Through the parameter identification of the discrete model of the power battery, the open circuit voltage Uoc of the power battery, the ohmic internal resistance R ₀ of the battery, the polarization internal resistance Rp and the polarization capacitance Cp of the battery are obtained.

Ⅰ-2第一变结构滤波的动力电池模型的参数辨识Ⅰ-2 Parameter identification of the power battery model of the first variable structure filter

采用变第一结构滤波对电池等效模型的离散模型进行参数辨识。根据公式(3)选择系统状态变量X_k＝[a_k,b_k,c_k,(1-a)U_oc,k]^T，得状态方程为：The parameters of the discrete model of the battery equivalent model are identified by variable first structure filtering. Select the system state variable X _k ＝[a _k ,b _k ,c _k ,(1-a)U _oc,k ] ^T according to formula (3), and the state equation is:

$\{\begin{matrix} {X x}_{k k} = = {X x}_{k k - - 11} + + {w w}_{k k - - 11} \\ {Z Z}_{k k} = = {U u}_{k k} = = {Cx Cx}_{k k} + + {v v}_{k k} \end{matrix} - - - - - - ((44))$

其中k是当前时刻，x_k∈R_n×1是系统状态向量；z_k∈R_m×1是测量状态变量；w_k和v_k分别为系统零均值随机过程噪声和测量噪声。主要由传感器精度、模型误差等造成；C是测量方程系数：C＝[U_k-1,I_k,I_k-1,1]。Where k is the current moment, x _k ∈ R _n ×1 is the system state vector; z _k ∈ R _m ×1 is the measurement state variable; w _k and v _k are the zero-mean random process noise and measurement noise of the system, respectively. It is mainly caused by sensor accuracy, model error, etc.; C is the measurement equation coefficient: C=[U _k-1 ,I _k ,I _k-1 ,1].

变结构滤波-Ⅰ算法下的参数辨识过程如下：The parameter identification process under variable structure filtering-I algorithm is as follows:

$[\begin{matrix} {\overset{^^}{z z}}_{k k | | k k - - 11} \\ {\overset{^^}{z z}}_{k k | | k k} \end{matrix}] = = C C [\begin{matrix} {\overset{^^}{X x}}_{k k | | k k - - 11} \\ {\overset{^^}{X x}}_{k k | | k k} \end{matrix}] - - - - - - ((55))$

${\overset{^^}{X x}}_{k k | | k k - - 11} = = {\overset{^^}{X x}}_{k k - - 11 | | k k - - 11} - - - - - - ((66))$

${\overset{^^}{X x}}_{k k | | k k} = = {\overset{^^}{X x}}_{k k | | k k - - 11} + + {K K}_{k k} - - - - - - ((77))$

第一变结构滤波的是系统状态变量X在k-1时刻的修正值，是系统状态变量X在k时刻的修正值，是系统状态变量X在k时刻的预测值；是测量状态变量z在k时刻的预测值；是测量状态变量z在k时刻的修正值；K_k是参数辨识在k时刻的修正增量，即系统状态变量X在k时刻的预测值的修正量。The first variable structure filter is the correction value of the system state variable X at time k-1, is the correction value of the system state variable X at time k, is the predicted value of the system state variable X at time k; is the predicted value of the measured state variable z at time k; is the correction value of the measured state variable z at time k; K _k is the correction increment of parameter identification at time k, that is, the predicted value of the system state variable X at time k correction amount.

第一变结构滤波通过调整变结构滤波的修正增量K_k，不断修正状态变量X在k时刻的预测值k时刻的修正增量K_k为：The first variable structure filter continuously corrects the predicted value of the state variable X at time k by adjusting the correction increment K _k of the variable structure filter The correction increment K _{k at time k} is:

$\{\begin{matrix} {e e}_{{z z}_{k k | | k k - - 11}} = = {z z}_{k k} - - {\overset{^^}{z z}}_{k k | | k k - - 11} \\ {e e}_{{z z}_{k k - - 11 | | k k - - 11}} = = {z z}_{k k - - 11} - - {\overset{^^}{z z}}_{k k - - 11 | | k k - - 11} \end{matrix} - - - - - - ((99))$

(8)、(9)式中z_k是电池运行时的端电压k时刻的测量值，即U_k；是当前k时刻测量状态变量z的真实值与k时刻预测值之间的误差；是k-1时刻测量状态变量的真实值与k-1时刻预测值之间的误差；C^-1是向量C的逆矩阵；β、γ是常数值，取值范围均为0～1，且β的取值直接影响变结构滤波参数识别的准确性；ο是Schur乘积，即两个矩阵对应元素相乘的结果；sat是饱和函数，其中Ψ为第一变结构滤波的平滑边界层厚度，第一变结构滤波的sat饱和函数的矢量定义如下：(8), (9) where z _k is the measured value of the terminal voltage k when the battery is running, that is, U _k ; is the error between the actual value of the measured state variable z at the current k time and the predicted value at the k time; is the error between the actual value of the measured state variable at time k-1 and the predicted value at time k-1; C ^-1 is the inverse matrix of vector C; β and γ are constant values, both of which range from 0 to 1, and The value of β directly affects the accuracy of variable structure filtering parameter identification; ο is the Schur product, that is, the result of multiplying the corresponding elements of two matrices; sat is a saturation function, where Ψ is the smooth boundary layer thickness of the first variable structure filter, The vector definition of the sat saturation function of the first variable structure filter is as follows:

$s the s a a t t (({e e}_{{z z}_{k k | | k k - - 11}},, ψ ψ)) = = {[\begin{matrix} s the s a a t t (({e e}_{{z z}_{11,, k k | | k k - - 11}},, {ψ ψ}_{11})) & ... ... & s the s a a t t (({e e}_{{z z}_{n no,, k k | | k k - - 11}},, {ψ ψ}_{n no})) \end{matrix}]}^{T T} - - - - - - ((1010))$

其中第一变结构滤波的饱和函数sat的定义如下：The saturation function sat of the first variable structure filter is defined as follows:

$s the s a a t t (({e e}_{{z z}_{i i,, k k | | k k - - 11}},, {ψ ψ}_{i i})) = = \{\begin{matrix} {e e}_{{z z}_{i i,, k k | | k k - - 11}} / / {ψ ψ}_{i i} & {e e}_{{z z}_{i i,, k k | | k k - - 11}} \leq \leq {ψ ψ}_{i i} \\ s the s i i g g n no (({e e}_{{z z}_{i i,, k k | | k k - - 11}})) & {e e}_{{z z}_{i i,, k k | | k k - - 11}} > > {ψ ψ}_{i i} \end{matrix} - - - - - - ((1111))$

其中Ψ_i是针对偏差为了引进边界层而给出的边界，|Ψ_i|为边界层的厚度，取为常量，|Ψ_i|＝0.01～0.03；sign表示符号函数：where Ψ _i is for the bias The boundary given for the introduction of the boundary layer, |Ψ _i | is the thickness of the boundary layer, taken as a constant, |Ψ _i |=0.01～0.03; sign represents the sign function:

$s the s i i g g n no (({e e}_{{z z}_{i i,, k k | | k k - - 11}})) = = \{\begin{matrix} {e e}_{{z z}_{i i,, k k | | k k - - 11}} > > 00 \\ {e e}_{{z z}_{i i,, k k | | k k - - 11}} = = 00 \\ {e e}_{{z z}_{i i,, k k | | k k - - 11}} < < 00 \end{matrix} . .$

通过以上第一变结构滤波，得到系统状态向量X_k＝[a_k,b_k,c_k,(1-a)U_oc,k]^T的估计值为从系统状态向量估计值得到动力电池模型的参数值：欧姆内阻R₀、极化内阻Rp、极化电容Cp及k时刻开路电压的估计值U_oc，k，即OCV。Through the above first variable structure filtering, the estimated value of the system state vector X _k =[a _k ,b _k ,c _k ,(1-a)U _oc,k ] ^T is The parameter values of the power battery model are obtained from the estimated value of the system state vector: the ohmic internal resistance R ₀ , the polarization internal resistance Rp, the polarization capacitance Cp, and the estimated value U _oc,k of the open circuit voltage at time k, namely OCV.

Ⅱ、动力电池开路电压OCV-SOC关系的拟合Ⅱ. Fitting of OCV-SOC relationship of power battery open circuit voltage

开路电压算法中，同类型的车载动力电池的开路电压OCV与SOC值有良好的一致性，OCV-SOC的高价多项式逼近拟合数学模型关系如下：In the open circuit voltage algorithm, the open circuit voltage OCV of the same type of vehicle power battery has a good consistency with the SOC value, and the relationship between the high-priced polynomial approximation and fitting mathematical model of OCV-SOC is as follows:

$\begin{matrix} {U u}_{o o c c,, k k} = = g g (({SOC SOC}_{k k})) \\ = = {h h}_{11} {SOC SOC}_{k k}^{88} + + {h h}_{22} {SOC SOC}_{k k}^{77} + + {h h}_{33} {SOC SOC}_{k k}^{66} + + {h h}_{44} {SOC SOC}_{k k}^{55} + + {h h}_{55} {SOC SOC}_{k k}^{44} \\ + + {h h}_{66} {SOC SOC}_{k k}^{33} + + {h h}_{77} {SOC SOC}_{k k}^{22} + + {h h}_{88} {SOC SOC}_{k k}^{11} + + {h h}_{99} \end{matrix} - - - - - - ((1212))$

式中：h₁～h₉为OCV-SOC高价多项式拟合下的系数，逼近拟合后得：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_k表示在DST(Dynamic StressTest动态应力测试)工况下采用高精度电流测量，按公认的SOC定义法得到的在k时刻电池剩余电量值。In the formula: h ₁ ~ h ₉ are coefficients under OCV-SOC high-priced polynomial fitting, after approximation fitting: h ₁ =2.10×10 ³ , h ₂ =-7.38×10 ³ , h ₃ =9.98×10 ³ , h ₄ =-6.23×10 ³ , h ₅ =1.40×10 ³ , h ₆ =3.26×10 ² , h ₇ =-2.40×10 ² , h ₈ =47.98, h ₉ =22.27. SOC _k represents the remaining battery power value at time k obtained by using high-precision current measurement under the DST (Dynamic Stress Test) working condition and according to the recognized SOC definition method.

Ⅲ、第二变结构滤波的SOC估算方法Ⅲ. The SOC estimation method of the second variable structure filter

采用第二变结构滤波进行SOC的在线估算，能有效保证SOC估算值的收敛性。选择SOC和电池模型中极化电容C_p的端电压U_p,k作为第二变结构滤波的系统状态变量，即XX_k＝[SOC_k U_p,k]^T，系统的状态方程和测量方程如下：Using the second variable structure filter to estimate the SOC online can effectively ensure the convergence of the estimated value of the SOC. Select the terminal voltage U _p,k of polarized capacitance C _p in the SOC and battery model as the system state variable of the second variable structure filter, that is, XX _k =[SOC _k U _p,k ] ^T , the state equation and measurement equation of the system as follows:

$\{\begin{matrix} {XX XX}_{k k} = = {AXX AXX}_{k k - - 11} + + {BI BI}_{k k - - 11} + + {ww ww}_{k k} \\ {ZZ ZZ}_{k k} = = {U u}_{k k} = = {U u}_{o o c c,, k k} - - {R R}_{00} {I I}_{k k} - - {U u}_{P P,, k k} + + {vv vv}_{k k} \end{matrix} - - - - - - ((1313))$

其中： in:

(13)式中：T为采样周期；Q_N为电池额定容量；η为充放电库伦效率；(13) In the formula: T is the sampling period; Q _N is the rated capacity of the battery; η is the coulombic efficiency of charging and discharging;

R_p表示电池的极化内阻；C_p表示电池的极化电容；U_oc,k表示k时刻电池的开路电压；R_o表示电池的欧姆内阻；I_k表示k时刻流过欧姆内阻R_o的电流；U_k表示电池运行时的k时刻端电压；zz_k是第二变结构滤波的测量状态变量，ww_k和vv_k分别为第二变结构滤波的系统零均值随机过程噪声和测量噪声，其方差与第一变结构滤波的系统零均值随机过程噪声w_k和测量噪声v_k的方差不同。主要由传感器精度、模型误差等造成。R _p represents the polarization internal resistance of the battery; C _p represents the polarization capacitance of the battery; U _oc,k represents the open circuit voltage of the battery at k time; R _o represents the ohmic internal resistance of the battery; I _k represents the ohmic internal resistance flowing through k time The current of R _o ; U _k represents the terminal voltage at time k when the battery is running; zz _k is the measured state variable of the second variable structure filter, ww _k and vv _k are the system zero-mean random process noise and A measurement noise whose variance is different from the variance of the first variable structure filtered systematic zero-mean stochastic process noise w _k and the measurement noise v _k . It is mainly caused by sensor accuracy and model error.

把公式(12)代入测量方程(13)整理得：Substitute formula (12) into measurement equation (13) to get:

$\begin{matrix} {U u}_{k k} = = {CCX CCX}_{k k} + + {h h}_{11} {SOC SOC}_{k k}^{88} + + {h h}_{22} {SOC SOC}_{k k}^{77} + + {h h}_{33} {SOC SOC}_{k k}^{66} + + {h h}_{44} {SOC SOC}_{k k}^{55} \\ + + {h h}_{55} {SOC SOC}_{k k}^{44} + + {h h}_{66} {SOC SOC}_{k k}^{33} + + {h h}_{77} {SOC SOC}_{k k}^{22} + + {h h}_{99} - - {R R}_{00} {I I}_{k k} \end{matrix} - - - - - - ((1414))$

其中CC为测量方程的系数，CC＝[h₈ -1]；SOC_k表示k时刻电池的剩余电量估算值。Among them, CC is the coefficient of the measurement equation, CC=[h ₈ -1]; SOC _k represents the estimated value of the remaining power of the battery at time k.

根据Ⅰ-2步骤中第一变结构滤波公式(7)得：According to the first variable structure filtering formula (7) in step I-2:

状态更新时中的KK_k为修正增量，即的修正值：when the status is updated KK _k in is the correction increment, namely Correction value for :

其中：是第二变结构滤波的系统状态变量xx在k时刻的修正值，是第二变结构滤波的系统状态变量XX在k时刻的预测值；CC^-1是向量CC的逆反矩阵；in: is the correction value of the system state variable xx of the second variable structure filtering at time k, is the predicted value of the system state variable XX of the second variable structure filtering at time k; CC ^-1 is the inverse matrix of the vector CC;

$\{\begin{matrix} {e e}_{{zz zz}_{k k / / k k - - 11}} = = Z Z {Z Z}_{k k} - - Z Z {\overset{^^}{Z Z}}_{k k / / k k - - 11} \\ {e e}_{{zz zz}_{k k - - 11 / / k k - - 11}} = = {ZZ ZZ}_{k k - - 11} - - Z Z {\overset{^^}{Z Z}}_{k k - - 11 / / k k - - 11} \end{matrix}$

是第二变结构滤波的k时刻测量系统状态变量的真实值与预测值之间的误差；是第二变结构滤波的k-1时刻测量系统状态变量的真实值与修正后的预测值之间的误差；ο是Schur乘积；sat是饱和函数，其中ψψ为第二变结构滤波的平滑边界层厚度，第二变结构滤波的sat饱和函数的矢量具体定义如下： is the error between the actual value and the predicted value of the measured system state variable at time k of the second variable structure filter; is the error between the real value of the state variable of the second variable structure filtering and the corrected predicted value at k-1 moment; o is the Schur product; sat is a saturation function, where ψψ is the smooth boundary of the second variable structure filtering Layer thickness, the specific definition of the vector of the sat saturation function of the second variable structure filtering is as follows:

$s the s a a t t (({e e}_{{zz zz}_{k k | | k k - - 11}},, ψ ψ ψ ψ)) = = {[\begin{matrix} s the s a a t t (({e e}_{{zz zz}_{11,, k k | | k k - - 11}},, {ψψ ψψ}_{11})) & ... ... & s the s a a t t (({e e}_{{zz zz}_{n no,, k k | | k k - - 11}},, {ψψ ψψ}_{n no})) \end{matrix}]}^{T T},,$

其中第二变结构滤波的饱和函数sat的定义如下：Wherein the definition of the saturation function sat of the second variable structure filtering is as follows:

$s the s a a t t (({e e}_{{zz zz}_{i i,, k k | | k k - - 11}},, {ψψ ψψ}_{i i})) = = \{\begin{matrix} {e e}_{{zz zz}_{i i,, k k | | k k - - 11}} / / {ψψ ψψ}_{i i} & {e e}_{{zz zz}_{i i,, k k | | k k - - 11}} \leq \leq {ψψ ψψ}_{i i} \\ s the s i i g g n no (({e e}_{{zz zz}_{i i,, k k | | k k - - 11}})) & {e e}_{{zz zz}_{i i,, k k | | k k - - 11}} > > {ψψ ψψ}_{i i} \end{matrix},,$

其中ψψ_i是针对偏差引进边界层而给出的边界，|ψψ_i|为边界层的厚度，取为常量，取值0.01～0.03；sign表示符号函数，其规则如下：where _ψψi is for the bias The boundary given by the introduction of the boundary layer, |ψψ _i | is the thickness of the boundary layer, which is taken as a constant value, ranging from 0.01 to 0.03; sign represents a sign function, and its rules are as follows:

$s the s i i g g n no (({e e}_{{zz zz}_{i i,, k k | | k k - - 11}})) = = \{\begin{matrix} {e e}_{{zz zz}_{i i,, k k | | k k - - 11}} > > 00 \\ {e e}_{{zz zz}_{i i,, k k | | k k - - 11}} = = 00 \\ {e e}_{{zz zz}_{i i,, k k | | k k - - 11}} < < 00 \end{matrix} . .$

(15)式的修正增量计算式中，ββ是第二变结构滤波的端电压估算值和端电压真实测量值之间误差的系数，γγ是第二变结构滤波的端电压修正值和端电压真实测量值之间误差的系数。In the modified incremental calculation formula of (15), ββ is the error between the estimated value of the terminal voltage of the second variable structure filter and the actual measured value of the terminal voltage The coefficient of , γγ is the error between the terminal voltage correction value of the second variable structure filter and the actual measurement value of the terminal voltage coefficient.

大量试验发现，ββ对修正增量KK_k值影响较大、且可控性强，本发明方法主要对参数ββ值进行调整。A large number of tests have found that ββ has a great influence on the correction increment KK _k value and is highly controllable. The method of the present invention mainly adjusts the parameter ββ value.

为了抑制系统在收敛阶段的抖振，本发明还引入模糊规则，针对变结构滤波在收敛后易产生振动，导致估算结果产生较大误差的特性，在变结构滤波进行SOC在线估算时，引入模糊规则调整变结构滤波参数ββ，增强修正增量的适应性。选取简洁而有效、既确保SOC精度又保证SOC收敛速度的ββ值，ββ取值的模糊规则如下：In order to suppress the chattering of the system during the convergence stage, the present invention also introduces fuzzy rules, aiming at the characteristics that the variable structure filter is prone to vibration after convergence, resulting in large errors in the estimation results, when the variable structure filter performs SOC online estimation, fuzzy rules are introduced. Regularly adjust variable structure filter parameters ββ to enhance the adaptability of correction increments. Select a simple and effective ββ value that not only ensures SOC accuracy but also ensures SOC convergence speed. The fuzzy rules for ββ value are as follows:

$β β β β = = \{\begin{matrix} 11 & {e e}_{{zz zz}_{k k / / k k - - 11}} &GreaterEqual; &Greater Equal; 0.034 0.034 {U u}_{N N} \\ 0.1 0.1 \cdot \cdot {e e}_{{zz zz}_{k k / / k k - - 11}} & 0.034 0.034 {U u}_{N N} > > {e e}_{{zz zz}_{k k / / k k - - 11}} > > - - 0.034 0.034 {U u}_{N N} \\ 11 & {e e}_{{zz zz}_{k k / / k k - - 11}} \leq \leq - - 0.034 0.034 {U u}_{N N} \end{matrix} - - - - - - ((1616))$

式中是测量系统状态变量的真实值与预测值之间的误差；U_N为电池的额定电压,本发明取12～144V.In the formula is the error between the actual value and the predicted value of the measurement system state variable; U _N is the rated voltage of the battery, which is 12-144V in the present invention.

通过式(16)对变结构滤波进行参数的自适应模糊调整，即为模糊变结构滤波。该模糊变结构滤波与用于电池参数辨识用的变结构滤波结合一起构成本发明的双变结构滤波的SOC估计方法，求得动力电池的SOC值。The self-adaptive fuzzy adjustment of the parameters of the variable structure filter by formula (16) is the fuzzy variable structure filter. The fuzzy variable structure filter is combined with the variable structure filter used for battery parameter identification to form the SOC estimation method of the double variable structure filter of the present invention to obtain the SOC value of the power battery.

本发明根据上述一种双变结构滤波的动力电池荷电状态的估计方法设计的双变结构滤波的动力电池荷电状态SOC的估计系统，包括微处理器、模数转换模块、电流传感器、电压传感器。According to the estimation method of the power battery state of charge of the power battery with the double variable structure filter, the estimation system of the state of charge of the power battery with the double variable structure filter is designed according to the present invention, including a microprocessor, an analog-to-digital conversion module, a current sensor, a voltage sensor.

电压传感器和电流传感器分别安装于待检测的动力电池端口，检测动力电池端电压和端口的电流。电压、电流传感器经模数转换模块连接微处理器，微处理器输出当前电池荷电状态的估计值。微处理器含有数据存储器和程序存储器，所述程序存储器内含有变结构滤波动力电池参数辨识模块，开路电压与SOC关系拟合模块以及参数自适应模糊调整模块。所述数据存储器存储动力电池的电压、电流检测数据和程序存储器中各模块计算过程中产生的数据。The voltage sensor and the current sensor are respectively installed at the port of the power battery to be detected to detect the terminal voltage of the power battery and the current of the port. The voltage and current sensors are connected to the microprocessor through the analog-to-digital conversion module, and the microprocessor outputs the estimated value of the current state of charge of the battery. The microprocessor includes a data memory and a program memory, and the program memory includes a variable structure filter power battery parameter identification module, an open circuit voltage and SOC relationship fitting module, and a parameter self-adaptive fuzzy adjustment module. The data memory stores the voltage and current detection data of the power battery and the data generated in the calculation process of each module in the program memory.

微处理器与显示屏连接，在线显示当前的电池荷电状态的估计值，即当前估计的SOC值。The microprocessor is connected with the display screen, and displays the estimated value of the current state of charge of the battery online, that is, the currently estimated SOC value.

微处理器配有CAN接口，以便与汽车的CAN控制器连接，即与汽车其他电气通信连接。The microprocessor is equipped with a CAN interface to connect with the car's CAN controller, that is, to connect with other electrical communications of the car.

电压传感器和电流传感器检测得到当前动力电池端电压和端口的电流的模拟信号经模数转换模块转换为对应的数字信号，送入微处理器，微处理器内的变结构滤波动力电池参数辨识模块、开路电压与SOC关系拟合模块以及参数自适应模糊调整模块，根据当前的端电压和电流值，按本发明的双变结构滤波的动力电池荷电状态的估计方法进行估算，得到当前的电池荷电状态的估计值。The voltage sensor and current sensor detect the current analog signal of the power battery terminal voltage and port current, which is converted into a corresponding digital signal by the analog-to-digital conversion module and sent to the microprocessor. The variable structure filter power battery parameter identification module in the microprocessor, The open-circuit voltage and SOC relationship fitting module and the parameter self-adaptive fuzzy adjustment module, according to the current terminal voltage and current value, estimate according to the method for estimating the state of charge of the power battery with double variable structure filtering in the present invention, and obtain the current battery charge An estimate of the electrical state.

与现有技术相比，本发明一种双变结构滤波的动力电池荷电状态的估计方法与系统优点为：1、变结构滤波的系统易于实现，结构简洁，可在线辨识动力电池参数；2、对变结构滤波进行参数修正增量进行自适应模糊调整，引入模糊控制规则，得到双变结构滤波的SOC估计方法；3、本估计方法简洁，运算量小，估计精度高，对SOC初值依赖性小；与常规变结构滤波的SOC估计相比，SOC的估计精度从原来的11.09％提高到1.50％。Compared with the prior art, the present invention has the advantages of a method and system for estimating the state of charge of a power battery with double variable structure filtering: 1. The system of variable structure filtering is easy to implement, has a simple structure, and can identify power battery parameters online; 2. 1. Carry out self-adaptive fuzzy adjustment on parameter correction increments of variable structure filtering, introduce fuzzy control rules, and obtain SOC estimation method of double variable structure filtering; 3. This estimation method is simple, with small calculation amount and high estimation accuracy. The dependence is small; compared with the SOC estimation of conventional variable structure filtering, the estimation accuracy of SOC is improved from 11.09% to 1.50%.

附图说明Description of drawings

图1为双变结构滤波的动力电池荷电状态的估计方法实施例的步骤Ⅰ-1所述的电池等效模型示意图；Fig. 1 is a schematic diagram of a battery equivalent model described in step I-1 of an embodiment of a method for estimating the state of charge of a power battery with double variable structure filtering;

图2为双变结构滤波的动力电池荷电状态的估计方法实施例的模糊变结构滤波与用于电池参数辨识用的变结构滤波的SOC估计方法示意图；Fig. 2 is a schematic diagram of the fuzzy variable structure filter and the SOC estimation method of the variable structure filter used for battery parameter identification in an embodiment of the method for estimating the state of charge of a power battery with double variable structure filtering;

图3为双变结构滤波的动力电池荷电状态的估计方法实施例的SOC估计流程示意图；Fig. 3 is a schematic diagram of the SOC estimation process embodiment of the method for estimating the state of charge of the power battery with double variable structure filtering;

图4为双变结构滤波的动力电池荷电状态的估计系统实施例的结构示意图。Fig. 4 is a schematic structural diagram of an embodiment of a system for estimating the state of charge of a power battery with double variable structure filtering.

具体实施方式detailed description

双变结构滤波的动力电池荷电状态SOC的估计方法实施例Embodiment of the estimation method of power battery state of charge SOC with double variable structure filter

本双变结构滤波的动力电池荷电状态SOC的估计方法实施例的流程图如图3所示，主要步骤如下：The flow chart of the embodiment of the method for estimating the power battery state of charge SOC of the double variable structure filter is shown in Figure 3, and the main steps are as follows:

Ⅰ、动力电池的参数辨识Ⅰ. Parameter identification of power battery

Ⅰ-1、动力电池离散模型Ⅰ-1. Discrete model of power battery

本例采用Thevenin模型为电池等效模型，描述电池的静态和动态性能，如图1所示，电池的极化内阻R_p与电池的极化电容C_p并联构成一阶RC结构，表示电池的极化反应，RC两端电压为U_p(t)表示电池端电压；串接欧姆电阻R₀，流过欧姆内阻R₀的电流为i(t)，Uoc(t)为电池的开路电压OCV，采样得到电池端电压U(t)和流过欧姆内阻R₀的电流i(t)。In this example, the _Thevenin _model is used as the battery equivalent model to describe the static and dynamic performance of the battery. As shown in Fig. Polarization reaction, the voltage across the RC is U _p (t) represents the battery terminal voltage; the ohmic resistance R ₀ is connected in series, the current flowing through the ohmic internal resistance R ₀ is i(t), and Uoc(t) is the open circuit of the battery The voltage OCV is sampled to obtain the battery terminal voltage U(t) and the current i(t) flowing through the ohmic internal resistance R ₀ .

本电池等效模型数学表达式如下：The mathematical expression of the battery equivalent model is as follows:

$\{\begin{matrix} \frac{{dU U}_{p p} ((t t))}{d d t t} = = - - \frac{{U u}_{p p} ((t t))}{{R R}_{p p} {C C}_{p p}} + + \frac{i i ((t t))}{{C C}_{p p}} \\ U u ((t t)) = = {U u}_{O o C C} ((t t)) - - {R R}_{00} i i ((t t)) - - {U u}_{p p} ((t t)) \end{matrix} - - - - - - ((11))$

对公式(2)整理得动力电池模型的离散模型为：According to the formula (2), the discrete model of the power battery model is:

${R R}_{00} = = \frac{{c c}_{k k}}{{a a}_{k k}},,$

式中T为采样周期，本例T为1秒。In the formula, T is the sampling period, and T is 1 second in this example.

其中k是当前时刻，X_k∈R_n×1是系统状态向量；Z_k∈R_m×1是测量状态变量；w_k和v_k分别为系统零均值随机过程噪声和测量噪声。C是测量方程系数，C＝[U_k-1,I_k,I_k-1,1]。Where k is the current moment, X _k ∈ R _n ×1 is the system state vector; Z _k ∈ R _m ×1 is the measurement state variable; w _k and v _k are the zero-mean random process noise and measurement noise of the system, respectively. C is the measurement equation coefficient, C=[U _k-1 , I _k , I _k-1 ,1].

OCV-SOC的高价多项式逼近拟合数学模型关系如下：The high-priced polynomial approximation fitting mathematical model relationship of OCV-SOC is as follows:

式中：h₁～h₉为OCV-SOC高价多项式拟合下的系数，逼近拟合后得：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_k表示在DST工况下采用高精度电流测量，按公认的SOC定义法得到的在k时刻电池剩余电量值。In the formula: h ₁ ~ h ₉ are coefficients under OCV-SOC high-priced polynomial fitting, after approximation fitting: h ₁ =2.10×10 ³ , h ₂ =-7.38×10 ³ , h ₃ =9.98×10 ³ , h ₄ =-6.23×10 ³ , h ₅ =1.40×10 ³ , h ₆ =3.26×10 ² , h ₇ =-2.40×10 ² , h ₈ =47.98, h ₉ =22.27. SOC _k represents the value of remaining battery power at time k obtained by using high-precision current measurement under DST working conditions and according to the recognized SOC definition method.

选择SOC和电池模型中极化电容C_p的端电压U_p,k作为第二变结构滤波的系统状态变量，即XX_k＝[SOC_k U_p,k]^T，系统的状态方程和测量方程如下：Select the terminal voltage U _p,k of polarized capacitance C _p in the SOC and battery model as the system state variable of the second variable structure filter, that is, XX _k =[SOC _k U _p,k ] ^T , the state equation and measurement equation of the system as follows:

其中： in:

(13)式中：T为采样周期；Q_N为电池额定容量；η为充放电库伦效率；R_p表示电池的极化内阻；C_p表示电池的极化电容；U_oc,k表示k时刻电池的开路电压；R_o表示电池的欧姆内阻；I_k表示k时刻流过欧姆内阻R_o的电流；U_k表示电池运行时的k时刻端电压；ZZ_k是第二变结构滤波的测量状态变量，ww_k和vv_k分别为第二变结构滤波的系统零均值随机过程噪声和测量噪声，其方差也与第一变结构滤波的系统零均值随机过程噪声w_k和测量噪声v_k的方差不同，主要由传感器精度、模型误差等造成。(13) where: T is the sampling period; Q _N is the rated capacity of the battery; η is the charge-discharge coulombic efficiency; R _p represents the polarization internal resistance of the battery; C _p represents the polarization capacitance of the battery; U _oc,k represents k The open circuit voltage of the battery at time; R _o represents the ohmic internal resistance of the battery; I _k represents the current flowing through the ohmic internal resistance R _o at time k; U _k represents the terminal voltage at time k when the battery is running; ZZ _k is the second variable structure filter The measurement state variables of , ww _k and vv _k are the system zero-mean random process noise and measurement noise of the second variable structure filter respectively, and their variances are also the same as the system zero-mean random process noise w _k and measurement noise v of the first variable structure filter The variance of _k is different, which is mainly caused by sensor accuracy, model error, etc.

其中CC为测量方程的系数，CC＝[h₈-1]；SOC_k表示k时刻电池的剩余电量估算值。Among them, CC is the coefficient of the measurement equation, CC=[h ₈ -1]; SOC _k represents the estimated value of the remaining power of the battery at time k.

是第二变结构滤波的k时刻测量系统状态变量的真实值与预测值之间的误差；是第二变结构滤波的k-1时刻测量系统状态变量的真实值与修正后的预测值之间的误差；ο是Schur乘积；sat是饱和函数，其中ψψ为第二变结构滤波的平滑边界层厚度第二变结构滤波的sat饱和函数的矢量具体定义如下： is the error between the actual value and the predicted value of the measured system state variable at time k of the second variable structure filter; is the error between the real value of the state variable of the second variable structure filtering and the corrected predicted value at k-1 moment; o is the Schur product; sat is a saturation function, where ψψ is the smooth boundary of the second variable structure filtering The specific definition of the vector of the sat saturation function of layer thickness second variable structure filtering is as follows:

其中ψψ_i是针对偏差引进边界层而给出的边界，|ψψ_i|为边界层的厚度，本例为0.02；sign表示符号函数，其规则如下：where _ψψi is for the bias The boundary given by introducing the boundary layer, |ψψ _i | is the thickness of the boundary layer, which is 0.02 in this example; sign represents the sign function, and its rules are as follows:

本例主要对参数ββ值进行调整。This example mainly adjusts the parameter ββ value.

ββ取值的模糊规则如下：The fuzzy rules for the value of ββ are as follows:

$β β β β = = \{\begin{matrix} 11 & {e e}_{{zz zz}_{k k / / k k - - 11}} &GreaterEqual; &Greater Equal; 0.034 0.034 {U u}_{N N} \\ 0.1 0.1 \cdot &Center Dot; {e e}_{{zz zz}_{k k / / k k - - 11}} & 0.034 0.034 {U u}_{N N} > > {e e}_{{zz zz}_{k k / / k k - - 11}} > > - - 0.034 0.034 {U u}_{N N} \\ 11 & {e e}_{{zz zz}_{k k / / k k - - 11}} \leq \leq - - 0.034 0.034 {U u}_{N N} \end{matrix} - - - - - - ((1616))$

式中是测量系统状态变量的真实值与预测值之间的误差；U_N为电池的额定电压,本例为24V。In the formula is the error between the actual value and the predicted value of the measurement system state variable; U _N is the rated voltage of the battery, which is 24V in this example.

通过式(16)对变结构滤波进行参数的自适应模糊调整，即为模糊变结构滤波。该模糊变结构滤波与用于电池参数辨识用的变结构滤波结合一起得到动力电池的SOC估计值，如图2所示。The self-adaptive fuzzy adjustment of the parameters of the variable structure filter by formula (16) is the fuzzy variable structure filter. The fuzzy variable structure filter is combined with the variable structure filter used for battery parameter identification to obtain the estimated value of SOC of the power battery, as shown in Fig. 2 .

双变结构滤波的动力电池荷电状态SOC的估计系统实施例Embodiment of the estimation system of power battery state of charge SOC with double variable structure filter

本双变结构滤波的动力电池荷电状态SOC的估计系统实施例，包括微处理器、模数转换模块、电流传感器、电压传感器。本例动力电池为24V，22Ah。The embodiment of the system for estimating the state of charge SOC of the power battery with dual variable structure filtering includes a microprocessor, an analog-to-digital conversion module, a current sensor, and a voltage sensor. In this example, the power battery is 24V, 22Ah.

电压传感器和电流传感器安装于动力电池端口，分别检测动力电池端电压和端口的电流。电压、电流传感器经模数转换模块连接微处理器的通用接口，微处理器输出端连接显示屏，在线显示当前的电池荷电状态的估计值，即当前估计的SOC值。微处理器输出端还连接CAN接口，与汽车的CAN控制器连接，即与汽车其他电气通信连接。The voltage sensor and the current sensor are installed at the port of the power battery to detect the terminal voltage of the power battery and the current of the port respectively. The voltage and current sensors are connected to the common interface of the microprocessor through the analog-to-digital conversion module, and the output terminal of the microprocessor is connected to the display screen, and the current estimated value of the battery state of charge, that is, the current estimated SOC value, is displayed online. The output end of the microprocessor is also connected to the CAN interface, and is connected with the CAN controller of the automobile, that is, connected with other electrical communications of the automobile.

微处理器含有数据存储器和程序存储器，所述程序存储器内含有变结构滤波动力电池参数辨识模块，开路电压与SOC关系拟合模块以及参数自适应模糊调整模块。数据存储器存储动力电池的电压、电流检测数据和程序存储器中各模块计算过程中产生的数据。The microprocessor includes a data memory and a program memory, and the program memory includes a variable structure filter power battery parameter identification module, an open circuit voltage and SOC relationship fitting module, and a parameter self-adaptive fuzzy adjustment module. The data memory stores the voltage and current detection data of the power battery and the data generated during the calculation of each module in the program memory.

电压传感器和电流传感器检测得到当前动力电池端电压和端口的电流的模拟信号经模数转换模块转换为对应的数字信号，送入微处理器，微处理器调用变结构滤波动力电池参数辨识模块、开路电压与SOC关系拟合模块以及参数自适应模糊调整模块，根据当前的端电压和电流值，按上述双变结构滤波的动力电池荷电状态的估计方法实施例进行估算，得到当前的电池荷电状态的估计值。The voltage sensor and current sensor detect the current power battery terminal voltage and port current. The analog signal is converted into a corresponding digital signal by the analog-to-digital conversion module and sent to the microprocessor. The microprocessor calls the variable structure filter power battery parameter identification module, open circuit The voltage and SOC relationship fitting module and the parameter self-adaptive fuzzy adjustment module estimate according to the above-mentioned embodiment of the method for estimating the state of charge of the power battery with double variable structure filtering according to the current terminal voltage and current value, and obtain the current battery charge Estimated value of the state.

上述实施例，仅为对本发明的目的、技术方案和有益效果进一步详细说明的具体个例，本发明并非限定于此。凡在本发明的公开的范围之内所做的任何修改、等同替换、改进等，均包含在本发明的保护范围之内。The above-mentioned embodiments are only specific examples for further specifying the purpose, technical solutions and beneficial effects of the present invention, and the present invention is not limited thereto. Any modifications, equivalent replacements, improvements, etc. made within the disclosed scope of the present invention are included in the protection scope of the present invention.

Claims

1. A method for estimating the SOC of a power battery with double variable structure filtering mainly comprises the following steps:

i, parameter identification of power battery

I-1 power battery discrete model

Describing the static and dynamic performances of the battery by adopting an equivalent model as a battery equivalent model; internal polarization resistance R of battery_pPolarization capacitance C with battery_pThe parallel connection forms a first-order RC structure which represents the polarization reaction of the battery, and the voltage at the two ends of the RC is U_p(t) represents a battery terminal voltage; in seriesMu resistor R₀Flowing through ohmic internal resistance R₀The current is i (t), Uoc (t) is the open circuit voltage OCV of the battery, and the terminal voltage U (t) and the ohmic internal resistance R flowing through the battery are obtained by sampling₀Current i (t);

the mathematical expression of the battery equivalent model is as follows:

\{\begin{matrix} \frac{{dU}_{p} (t)}{d t} = - \frac{U_{p} (t)}{R_{p} C_{p}} + \frac{i (t)}{C_{p}} \\ U (t) = U_{O C} (t) - R_{0} i (t) - U_{p} (t) \end{matrix} - - - (1)

discretizing the battery model of the formula (1) by adopting a backward difference transformation method to obtain:

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

wherein k is the current time, k-1 is the previous time, a, b and c are model parameters:

the discrete model of the power battery equivalent model obtained by the formula (2) is as follows:

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_kThe parameter relationship with the battery backward difference model is as follows:

R_{0} = \frac{c_{k}}{a_{k}},

R_{p} = \frac{- a_{k} b_{k} - c_{k}}{a_{k} (1 - a_{k})},

C_{p} = \frac{{Ta}_{k}^{2}}{- a_{k} b_{k} - c_{k}},

wherein T is a sampling period and is 0.5 to 2 seconds;

obtaining the parameters of the power battery, namely the open-circuit voltage Uoc and the ohmic internal resistance R of the battery through parameter identification of a discrete model of the power battery₀Polarization internal resistance Rp and polarization capacitance Cp of the battery;

parameter identification of I-2 first variable structure filtering power battery model

Performing parameter identification on a discrete model of the battery equivalent model by adopting variable first structure filtering; selecting a system state variable X according to equation (3)_k＝[a_k,b_k,c_k,(1-a)U_oc,k]^TThe state equation is obtained as follows:

\{\begin{matrix} X_{k} = X_{k - 1} + w_{k - 1} \\ Z_{k} = U_{k} = {Cx}_{k} + v_{k} \end{matrix} - - - (4)

where k is the current time, X_k∈R_n× 1 is the system state vector, Z_k∈R_m× 1 is a measured state variable, w_kAnd v_kRespectively representing system zero-mean random process noise and measurement noise; c is the coefficient of the measurement equation, C ═ U_k-1,I_k,I_k-1,1]；

The parameter identification process under the variable structure filter-I algorithm is as follows:

[\begin{matrix} {\hat{z}}_{k | k - 1} \\ {\hat{z}}_{k | k} \end{matrix}] = C [\begin{matrix} {\hat{X}}_{k | k - 1} \\ {\hat{X}}_{k | k} \end{matrix}] - - - (5)

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

{\hat{X}}_{k | k} = {\hat{X}}_{k | k - 1} + K_{k} - - - (7)

with filtering of a first variable structureIs a correction value of the system state variable X at the time k-1,is a correction value of the system state variable X at time k,is the predicted value of the system state variable X at the moment k;the predicted value of the state variable z at the moment k is measured;is a correction value of the measured state variable z at the time k; k_kIs the correction increment of parameter identification at the time k, namely the predicted value of the system state variable X at the time kThe correction amount of (1);

the first variable structure filtering is realized by adjusting the correction increment K of the variable structure filtering_kContinuously correcting the predicted value of the state variable X at the moment kCorrection increment K at time K_kComprises the following steps:

\{\begin{matrix} 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{matrix} - - - (9)

(8) and (9) wherein z is_kIs a measure of the terminal voltage k instant at which the battery is operating, i.e. U_k；Is between the true value of the current k moment measurement state variable z and the predicted value of the k momentAn error of (2);the error between the real value of the measured state variable at the moment k-1 and the predicted value at the moment k-1 is determined; c^-1The method comprises the following steps of obtaining a vector C, wherein β and gamma are constant values, the value ranges are 0-1, the value of β directly influences the accuracy of variable structure filtering parameter identification, the value is a Schur product, namely a multiplication result of corresponding elements of the two matrices, sat is a saturation function, psi is the thickness of a smooth boundary layer of first variable structure filtering, and the vector of the sat saturation function of the first variable structure filtering is defined as follows:

s a t (e_{z_{k | k - 1}}, ψ) = {[\begin{matrix} s a t (e_{z_{1, k | k - 1}}, ψ_{1}) & ... & s a t (e_{z_{n, k | k - 1}}, ψ_{n}) \end{matrix}]}^{T} - - - (10)

wherein the saturation function sat of the first variable structure filtering is defined as follows:

s a t (e_{z_{i},_{k | k - 1}}, ψ_{i}) = \{\begin{matrix} e_{z_{i, k | k - 1}} / ψ_{i} & e_{z_{i, k | k - 1}} \leq ψ_{i} \\ s i g n (e_{z_{i, k | k - 1}}) & e_{z_{i, k | k - 1}} > ψ_{i} \end{matrix} - - - (11)

therein Ψ_iIs directed to the deviationBoundary given for the introduction of boundary layer, | Ψ_iL is the thickness of the boundary layer, taken as a constant, | Ψ_i0.01-0.03, |; sign denotes the sign function:

s i g n (e_{z_{i, k | k - 1}}) = \{\begin{matrix} e_{z_{i, k | k - 1}} > 0 \\ e_{z_{i, k | k - 1}} = 0 \\ e_{z_{i, k | k - 1}} < 0 \end{matrix};

obtaining a system state vector X through the first variable structure filtering_k＝[a_k,b_k,c_k,(1-a)U_oc,k]^TIs estimated asObtaining parameter values of the power battery model from the system state vector estimation value: ohmic internal resistance R₀Polarization internal resistance Rp, polarization capacitance Cp and estimated value U of open-circuit voltage at moment k_oc，kI.e., OCV;

II, fitting of open-circuit voltage OCV-SOC relation of power battery

The high-price polynomial approximation fitting mathematical model relation of OCV-SOC is as follows:

\begin{matrix} U_{o c, k} = g ({SOC}_{k}) \\ = h_{1} {SOC}_{k}^{8} + h_{2} {SOC}_{k}^{7} + h_{3} {SOC}_{k}^{6} + h_{4} {SOC}_{k}^{5} + h_{5} {SOC}_{k}^{4} \\ + h_{6} {SOC}_{k}^{3} + h_{7} {SOC}_{k}^{2} + h_{8} {SOC}_{k}^{1} + h_{9} \end{matrix} - - - (12)

in the formula: h is₁～h₉And (3) approximating a coefficient under fitting of an OCV-SOC high-order polynomial to obtain: h is₁＝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_kRepresenting the value of the remaining battery power at the moment k;

III, SOC estimation method of second variable structure filtering

Selection of polarization capacitance C in SOC and Battery model_pTerminal voltage U of_p,kAs a system state variable for second variable structure filtering, i.e. XX_k＝[SOC_kU_p,k]^TThe system's equation of state and measurement equations are as follows:

\{\begin{matrix} {XX}_{k} = {AXX}_{k - 1} + {BI}_{k - 1} + {ww}_{k} \\ {ZZ}_{k} = U_{k} = U_{o c, k} - R_{0} I_{k} - U_{P, k} + {vv}_{k} \end{matrix} - - - (13)

wherein:

(13) in the formula: t is a sampling period; q_NRated capacity of battery, η coulombic efficiency of charge and discharge, R_pRepresenting the polarization internal resistance of the battery; c_pRepresents the polarization capacitance of the cell; u shape_oc,kRepresents the open circuit voltage of the battery at the moment k; r_oRepresents the ohmic internal resistance of the battery; i is_kIndicates the ohmic internal resistance R flowing at the time of k_oThe current of (a); u shape_kRepresenting terminal voltage at time k when the battery is running; zz (z)_kIs the measured state variable, ww, of the second variable structure filtering_kAnd vv_kRespectively carrying out zero mean random process noise and measurement noise of the system filtered by the second variable structure;

substituting equation (12) into measurement equation (13) to obtain:

\begin{matrix} U_{k} = {CCX}_{k} + h_{1} {SOC}_{k}^{8} + h_{2} {SOC}_{k}^{7} + h_{3} {SOC}_{k}^{6} + h_{4} {SOC}_{k}^{5} \\ + h_{5} {SOC}_{k}^{4} + h_{6} {SOC}_{k}^{3} + h_{7} {SOC}_{k}^{2} + h_{9} - R_{0} I_{k} \end{matrix} - - - (14)

where CC is the coefficient of the measurement equation, [ h ]₈-1]；SOC_kRepresenting the estimated value of the residual capacity of the battery at the moment k;

the first variable structure filtering formula (7) in the step I-2 is obtained:

when the status is updatedKK in (1)_kTo correct for increments, i.e.Correction value of (2):

wherein:is the corrected value of the system state variable XX of the second variable structure filtering at the moment k,is systematic form of second variable structure filteringThe predicted value of the state variable XX at the moment k; CC (challenge collapsar)^-1Is the inverse matrix of the vector CC;

\{\begin{matrix} e_{{zz}_{k / k - 1}} = {ZZ}_{k} - Z {\hat{Z}}_{k / k - 1} \\ e_{{zz}_{k - 1 / k - 1}} = {ZZ}_{k - 1} - Z {\hat{Z}}_{k - 1 / k - 1} \end{matrix}

the error between the real value and the predicted value of the state variable of the measurement system at the moment k of the second variable structure filtering is obtained;the error between the true value of the state variable of the measurement system at the k-1 moment of the second variable structure filtering and the corrected predicted value is obtained; omicron is the Schur product; sat is the saturation function, where ψ ψ is the smooth boundary layer thickness of the second variable structure filter the vector of the sat saturation function of the second variable structure filter is specifically defined as follows:

s a t (e_{{zz}_{k | k - 1}}, ψ ψ) = {[\begin{matrix} s a t (e_{{zz}_{1, k | k - 1}}, {ψψ}_{1}) & ... & s a t (e_{{zz}_{n, k | k - 1}}, {ψψ}_{n}) \end{matrix}]}^{T},

wherein the saturation function sat of the second variable structure filtering is defined as follows:

s a t (e_{{zz}_{i},_{k | k - 1}}, {ψψ}_{i}) = \{\begin{matrix} e_{{zz}_{i, k | k - 1}} / {ψψ}_{i} & e_{{zz}_{i, k | k - 1}} \leq {ψψ}_{i} \\ s i g n (e_{{zz}_{i, k | k - 1}}) & e_{{zz}_{i, k | k - 1}} > {ψψ}_{i} \end{matrix},

wherein psi_iIs directed to the deviationBoundary, | ψ given by introduction of boundary layer_iTaking the thickness of the boundary layer as a constant; sign denotes a sign function, which is regulated as follows:

s i g n (e_{{zz}_{i, k | k - 1}}) = \{\begin{matrix} e_{{zz}_{i, k | k - 1}} > 0 \\ e_{{zz}_{i, k | k - 1}} = 0 \\ e_{{zz}_{i, k | k - 1}} < 0 \end{matrix}

(15) in the modified incremental calculation of equation (III), ββ is the second variable structure filterError between estimated value of terminal voltage and true measured value of terminal voltageGamma is the error between the corrected value of the second variable structure filter and the real measured value of the terminal voltageThe coefficient of (a).

2. The estimation method of the SOC of the power battery with the double variable structure filtering according to claim 1, wherein:

the sampling period T is 0.5 to 2 seconds.

3. The estimation method of the SOC of the power battery with the double variable structure filtering according to claim 2, wherein:

the fuzzy rule of the beta value in the step III is as follows:

β β = \{\begin{matrix} 1 & e_{{zz}_{k / k - 1}} &GreaterEqual; 0.034 U_{N} \\ 0.1 \cdot e_{{zz}_{k / k - 1}} & 0.034 U_{N} > e_{{zz}_{k / k - 1}} > - 0.034 U_{N} \\ 1 & e_{{zz}_{k / k - 1}} \leq - 0.034 U_{N} \end{matrix} - - - (16)

in the formulaMeasuring the error between the real value and the predicted value of the system state variable; u shape_NIs the rated voltage of the battery.

4. The estimation method of the state of charge (SOC) of the double variable structure filtered power battery according to any one of claims 1 to 3, characterized in that:

the | ψ in said step III_iAnd | takes a value of 0.01-0.03.

5. The estimation system of the state of charge of the power battery with the double variable structure filtering, which is designed according to the estimation method of the state of charge SOC of the power battery with the double variable structure filtering, comprises a microprocessor, an analog-to-digital conversion module, a current sensor and a voltage sensor; the voltage sensor and the current sensor are respectively arranged at a port of a power battery to be detected, the voltage and the current of the port of the power battery are detected, the voltage and the current sensors are connected with the microprocessor through the analog-to-digital conversion module, and the microprocessor outputs an estimated value of the current battery charge state; the method is characterized in that:

the microprocessor comprises a data memory and a program memory, wherein the program memory comprises a variable structure filter power battery parameter identification module, an open-circuit voltage and SOC relation fitting module and a parameter self-adaptive fuzzy adjustment module; the data memory stores the voltage and current detection data of the power battery and data generated in the calculation process of each module in the program memory.

6. The system for estimating the state of charge (SOC) of a power battery with double variable structure filtering according to claim 5, wherein:

and the microprocessor is connected with the display screen and displays the estimated value of the current battery charge state on line.

7. The system for estimating the state of charge (SOC) of a power battery with double variable structure filtering according to claim 5, wherein:

the microprocessor is provided with a CAN interface.