CN102289557A - A battery model parameters and the remaining capacity estimation method joint asynchronous online - Google Patents

A battery model parameters and the remaining capacity estimation method joint asynchronous online Download PDF

Info

Publication number
CN102289557A
CN102289557A CN 201110127479 CN201110127479A CN102289557A CN 102289557 A CN102289557 A CN 102289557A CN 201110127479 CN201110127479 CN 201110127479 CN 201110127479 A CN201110127479 A CN 201110127479A CN 102289557 A CN102289557 A CN 102289557A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
battery
estimation
matrix
state
discharge
Prior art date
Application number
CN 201110127479
Other languages
Chinese (zh)
Other versions
CN102289557B (en )
Inventor
何志伟
曾毓
高明煜
黄继业
Original Assignee
杭州电子科技大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention relates to a battery model parameter and residual battery capacity joint asynchronous online estimation method. The existing method generally supposes that the parameters such as internal resistance of the same type of batteries is basically unchanged, thus the influence of battery aging on the residual battery capacity estimation precision is difficult to overcome. The method provided by the invention measures the battery end voltage and battery supply current at the current moment, estimates the residual battery capacity at the moment by a sampling point Kalman filtering algorithm based on the estimation result of moment battery model parameter according to a reasonable battery model and based on proper initialization, and then finishes estimation on the battery model parameter by the sampling point Kalman filtering algorithm by use of the residual battery capacity estimated at the moment. Estimation of the residual battery capacity and battery model parameter is asynchronously and alternately finished in an online mode. The method provided by the invention can conveniently perform online estimation on the residual battery capacity, has high convergence rate and high estimation precision, and suffers little influence of the battery aging.

Description

一种电池模型参数与剩余电量联合异步在线估计方法 A battery model parameters and the remaining capacity estimation method joint asynchronous online

技术领域 FIELD

[0001 ] 本发明属于电池技术领域,具体涉及一种电池模型参数与剩余电量联合异步在线估计方法。 [0001] The present invention belongs to the field of battery technology, particularly relates to a battery remaining power of the model parameters and the asynchronous line Joint Estimation method.

背景技术 Background technique

[0002] 电池作为备用电源已在通讯、电力系统、军事装备等领域得到了广泛的应用。 [0002] The battery as a backup power source has been widely used in the field of communications, power systems, military equipment. 同传统燃油汽车相比,电动汽车可实现零排放,因此是未来汽车的主要发展方向。 Compared with traditional fuel vehicles, electric vehicles can achieve zero emissions, and therefore the future of the main development direction of the car. 在电动汽车中电池直接作为主动能量供给部件,因此其工作状态的好坏直接关系到整个汽车的行驶安全性和运行可靠性。 Supplied directly to the battery in an electric vehicle as an active member of energy, and therefore its quality is directly related to the operating state of the entire car with the safety and operational reliability. 为确保电动汽车中的电池组性能良好,延长电池组使用寿命,须及时、准确地了解电池的运行状态,对电池进行合理有效的管理和控制。 To ensure good performance electric vehicle battery packs, extending battery life, to be timely and accurate understanding of the state of the battery, the battery is reasonable and effective management and control.

[0003] 电池荷电状态(State of Charge,以下简称S0C)的精确估算是电池能量管理系统中最核心的技术。 [0003] The battery state of charge (State of Charge, hereafter referred S0C) accurate estimation of battery energy management system is the core technology. 电池的SOC无法用一种传感器直接测得,它必须通过对一些其他物理量的测量,并采用一定的数学模型和算法来估计得到。 SOC of the battery can not be directly measured by a sensor which must by measuring other physical quantities, and the use of certain mathematical models and algorithms to estimate obtained.

[0004] 目前常用的电池SOC估计方法有开路电压法、安时法等。 [0004] The most commonly used method is the battery open circuit voltage SOC estimation method, when the security method and the like. 开路电压法进行电池SOC 估计时电池必须静置较长时间以达到稳定状态,而且只适用于电动汽车在停车状态下的SOC估计,不能满足在线检测要求。 When the battery SOC estimate open circuit voltage of the battery must stand for a long time to reach steady state, and apply only to electric cars in the parking SOC estimation state can not meet online testing requirements. 安时法易受到电流测量精度的影响,在高温或电流波动剧烈情况下,精度很差。 Method Ah susceptible to the measurement accuracy, at a high temperature or current volatility, the poor accuracy. 另一方面,已有方法一般都假设同类型的电池其内阻等参数基本不变,从而对同一类型电池进行SOC估计时均采用同一组模型参数,这种假设在电池没有发生老化时往往是成立的,但是当电池老化较严重时,电池内阻等会发生较大的变化,此时再基于原有模型参数进行SOC估计势必会发生较大程度的偏差。 On the other hand, a method has been generally assumed that the same type of battery and other parameters substantially constant internal resistance, so that the battery of the same type are used the same set of model parameters estimated SOC, this assumption does not occur when the battery is often aged established, but more serious when the battery is aging, the internal resistance of the battery and so will the big change, then re-SOC based on the original estimate model parameters bound to a greater degree of deviation occurs.

发明内容 SUMMARY

[0005] 本发明的目的就是克服现有技术的不足,提出一种电池模型参数与剩余电量联合异步在线估计方法,在在线估计出电池SOC的同时,可以对模型参数进行联合异步在线估计,从而克服由于电池老化带来的电池参数变化对电池SOC估计准确性的影响。 Objective [0005] The present invention is to overcome the disadvantages of the prior art, and provide a battery remaining power model parameter estimation methods combined asynchronous line, in-line, while the estimated SOC of the battery can be combined asynchronous online estimation of model parameters, such overcome the battery cell parameter changes arising from an aging effect on the estimation accuracy of the battery SOC. 本发明方法可以适用于所有电池,且估计精度较高。 The method of the present invention can be applied to all batteries and high estimation accuracy.

[0006] 本发明的电池模型参数与剩余电量联合异步在线估计方法,具体步骤是: [0006] the battery model and the parameters of the invention remaining amount asynchronous line Joint estimation method, the specific steps are:

步骤(1)测量在 A时刻的电池 Step (1) A measurement cell at time

端电压Λ和电池供电电流, t = 1,2, ι…。 Λ battery terminal voltage and the supply current, t = 1,2, ι ....

[0007] 步骤(¾用状态方程和观测方程表示电池的各个时刻的荷电状态依赖关系: [0007] Step (¾ a state equation and the observation equation represents the state of charge of the battery at each time dependence:

f f

状态方程:= /(¾,½) + Wa =¾- —- + Wfc State equation: = / (¾, ½) + Wa = ¾- - + Wfc

κ j κ j

6 6

Figure CN102289557AD00071

观测方程: Observation equation:

Figure CN102289557AD00072

其中z为电池的荷电状态,即剩余电量A为电池的放电比例系数,反映的是放电速率、温度等因素对电池soc的影响程度,本发明中只考虑放电速率的影响;a是电池在室温25 条件下、以1/30倍额定电流的放电速率放电时所能得到的额定总电量Δ是测量时间间隔,%为处理噪声。 Wherein z is the state of charge of the battery, i.e., the remaining amount A is a discharge ratio coefficient cells, reflects the degree of influence discharge rate, temperature and other factors on the battery soc, the present invention considers only affect the rate of discharge; a is battery 25 at room temperature conditions, a discharge rate of 30 times the rated current that can be obtained when discharging the rated total power is a measure of the time interval Δ,% to process noise. A= [^o R ^2 ^ ^4 f为电池观测模型的参数,是一个列向量;Λ为电池的内阻,、为观测噪声。 A = [^ o R ^ 2 ^ ^ 4 f is the observation model parameters of the battery, is a column vector; Lambda is the internal resistance of the battery ,, the observation noise.

[0008] 放电比例系数访的确定方法为: The method of determining the [0008] discharge ratio coefficient to visit:

(a)将完全充满电的电池以不同放电速率 (A) The fully charged battery at different discharge rates

Figure CN102289557AD00073

,c为电池的额定放电电流) , C is the nominal battery discharge current)

恒流放电-v(mo)次,计算相应放电速率下的电池总电量β , ι化N。 Constant discharge -v (mo), with the total battery β calculated in the respective discharge rate, of iota N.

[0009] (b)根据最小二乘方法拟合出β与Ci间的二次曲线关系,即在最小均方误差准则下求出同时满足 [0009] (b) fitting a quadratic curve the relationship between β and Ci, i.e., determined at meet minimum mean square error criterion of least squares method

Figure CN102289557AD00074

, aAc为最优系数。 , AAc optimal coefficients.

[0010] (C)在放电电流为s时,对应的放电比例系数"力: [0010] (C) when the discharge current is s, the discharge ratio coefficient corresponding to "power:

Figure CN102289557AD00075

此处,由于放电比例系数与电池老化等无关,因此,最优系数《Ac对于同一类型的电 Ac power Here, since the discharge ratio coefficient irrespective of aging of the battery, and therefore, the optimal coefficient "for the same type of

池只需确定一次,确定后可作为已知常数直接用于所有同类型电池的剩余电量估计。 Determining the pool only once, after the determination may be directly used as a known constant for all the remaining power of the battery of the same type estimation. [0011] 步骤(¾执行如下初始化过程: (a)电池剩余电量估计的初始化: [0011] Step (¾ to perform the initialization process: (a) initialization of the estimated remaining battery capacity:

起始状态為及其方差石分别为: Its initial state is the variance of the stone are:

Figure CN102289557AD00076

处理噪声巧的方差4、观测噪声H的方差\分别为: Coincidentally handling noise variance 4, the observation noise variance of H \ are:

Figure CN102289557AD00077

尺度参数7为: Scale parameter 7:

Figure CN102289557AD00078

扩展后的状态向量驾及其协方差if为: The extended state vector and covariance if to drive:

Figure CN102289557AD00079

均值加权系数w(m)= 0,1,2,...,6和方差加权系数w(m)= 0,1,2,...,6分别为 Mean weighting coefficient w (m) = 0,1,2, ..., 6, and variance weighting coefficient w (m) = 0,1,2, ..., 6 respectively,

Figure CN102289557AD00081

(b)电池模型参数估计的初始化: 任意选取初始模型参数PO=PO (B) initialization of the battery model parameter estimation: arbitrarily selecting initial model parameters PO = PO

设定PO的平方根均方差矩阵为SPO, SPO = I6 ;其中/6为6><6的单位矩阵; 选取比例常数A , A>1 ; PO is set to the square root of the mean square error matrix SPO, SPO = I6; wherein / 6 6> <6 unit matrix; Select proportionality constant A, A> 1;

设定变量民-/1Λ_3 Set the variable Man - / 1Λ_3

设定加权系数甙 Setting the weighting coefficients glycosides

Figure CN102289557AD00082

[0012] 步骤(4)采用采样点卡尔曼滤波算法进行循环递推: [0012] Step (4) The sample point Kalman filter algorithm Recursive:

在时刻λ = …,根据测得的电池端电圧H及电池的供电电流^ ,按下列步骤迭代 At time λ = ..., according to the supply current measured battery terminal electrically-pressure battery and H ^, the following iterative steps

进行电池模型参数与剩余电量的联合异步估计: (a)电池剩余电量的估计流程 And the battery remaining amount of the model parameter estimation combined asynchronous: estimating Process (a) of remaining battery power

①根据λ — 1时刻的扩展状态向量5^1及其协方差计算该时刻的所有的采样点序列礼: ① The λ - 1 calculates the time-expanded state vector and covariance 5 ^ 1 all the sample points of the time sequence of Li:

②根据状态方程进行时间域更新: ② updated according to the time-domain equation of state:

由采样点序列Il4,根据状态方程计算采样点更新3¾^ :¾-!=- From the sampling point sequence Il4, sampling point is calculated according to the equation of state update 3¾ ^:! ¾ - = -

对采样点更新进行加权,计算状态估计ρ Weighted sampling point update, the state estimation calculation ρ

计算状态估计乡-的方差 Calculation of state estimation town - variance

Figure CN102289557AD00083

③根据观测方程完成测量更新: ③ The measurement update complete observation equation:

由采样点更新3¾«及时刻的参数估计值iw,根据观测方程计算测量更新^ijw Updated by the sampling points 3¾ «time parameter estimates and iw, the observations measurement update equation ^ ijw

对测量更新m进行加权,计算测量估计y -Ji=ZMsh^i 计算测量估计的方差P- -.Ph=Z4'1 -λΓ M weighted measurement update, calculated measurement evaluation y -Ji = ZMsh ^ i calculated estimated variance of the measurement P- -.Ph = Z4'1 -λΓ

计算炉与尔的互协方差ρ : Calculation furnace and the cross-covariance Seoul ρ:

Figure CN102289557AD00084

计算卡尔曼增益 Kalman gain calculation

Figure CN102289557AD00091

计算状态更新 Calculation status update

Figure CN102289557AD00092

计算状态更新4的方差 Variance calculation status update 4

Figure CN102289557AD00093

通过上述流程,所得到的状态更新值4即为当前时刻i所估计得到的电池剩余电量。 Through the above process, the resultant state is the updated value of the current time i 4 are obtained estimated remaining battery capacity.

[0013] (b)电池模型参数的估计流程: ①计算模型参数的估计值 Estimation Process [0013] (b) the battery model parameters: the estimated value of the model parameters ①

Figure CN102289557AD00094

计算模型参数的平方根均方差矩阵的估计值 Calculate the square root of the model parameters are estimated value of the covariance matrix

Figure CN102289557AD00095

,其中, ,among them,

Figure CN102289557AD00096

为对应矩阵的对角线元素构成的列向量。 Corresponding column vector of the diagonal elements of the matrix configuration.

[0014] ②计算芮的采样点序列Xiw : [0014] ② calculated sampling points Rui sequence Xiw:

Figure CN102289557AD00097

P;为6 X 1列向量,ίς为6 X 6矩阵,故Xlw为6 X 13矩阵。 P; 6 X 1 is a column vector, ίς to 6 X 6 matrix, so Xlw of 6 X 13 matrix.

[0015] ③按下列各式计算测量更新: [0015] ③ measurement update calculated by the following formulas:

计算采样点的观测序列 Observation sequence calculating sample points

Figure CN102289557AD00098

6X13矩阵; 6X13 matrix;

计算观测序列m,的估计值 Estimated value calculating observation sequence m, the

Figure CN102289557AD00099

的第.列; . The first column;

Figure CN102289557AD000910

计算观测序列5&«的平方根均方差矩阵巧 Calculating observation sequence 5 & «square root of the covariance matrix are clever

Figure CN102289557AD000911

计算协方差矩阵 Covariance matrix

Figure CN102289557AD000912

计算卡尔曼增益:^ Kalman gain is calculated: ^

Figure CN102289557AD000913

计算参数更新 Calculation parameters Update

Figure CN102289557AD000914

计算临时变量t/ :u = Klsit ; Calculating a temporary variable t /: u = Klsit;

计算模型参数的平方根均方差矩阵的更新Sit -.Sfi=ChdS^JJrXj , 其中表示求矩阵的正交三角分解,并返回得到的上三角矩阵;Cf为矩阵的转置操作.'ch—daiM.S-?t,U,-\、表示求矩阵(¾/ *S;t-Umr 的Cholesky 分解。 Square-root calculation of the model parameters are updated covariance matrix Sit -.Sfi = ChdS ^ JJrXj, which represents an orthogonal triangular decomposition of Matrix, and returns the upper triangular matrix obtained; of Cf is a transpose operation .'ch-daiM matrix. ? S- t, U, - \, represents the matrix (¾ / * S; t-Umr the Cholesky decomposition.

[0016] 通过上述流程,所得到的良即为当前时刻i所估计得到的电池模型参数。 [0016] Through the above process, the resulting good is the current time i by the battery model parameter estimation obtained.

[0017] 在每一时刻,上述步骤4 (a)、4(b)交替进行,因此,电池剩余电量的估计依赖于上一时刻电池模型参数的估计结果,另一方面,电池模型参数的估计则基于当前时刻所估计得到的电池剩余电量完成。 [0017] At each time point, the above step 4 (a), 4 (b) alternately, and therefore, the estimated remaining battery power is estimated dependency estimation result of battery model parameters on one time, on the other hand, the battery model parameters remaining battery power based on the estimated current time is completed. 整个循环递推过程是在线完成的,即在电池实际工作过程中在线异步完成各时刻电池剩余电量的估计与电池模型参数的估计。 Recursive whole process is done online, i.e. during the actual cell line asynchronously estimation and model parameters each time the battery remaining battery capacity.

[0018] 本发明可以方便地进行电池SOC的快速估计,且可以克服电池老化对模型参数的影响。 [0018] The present invention can be quickly and easily estimate the battery SOC, and the battery can overcome the effects of aging on the model parameters. 该方法收敛速度快,估计精度高,而且适用于各种电池SOC的快速估计。 The method converges fast, accurate estimation, but also for rapid evaluation of various battery SOC.

[0019] 根据本发明的第一方面,公开了一种用于电池模型参数与剩余电量联合异步在线估计方法所依赖的测量量,分别为电池的端电压和电池的供电电流。 [0019] According to a first aspect of the present invention, discloses measuring a battery remaining amount and the model parameters for the asynchronous line Joint estimation methods rely on, respectively, the supply current of the battery and the terminal voltage of the battery.

[0020] 根据本发明的第二方面,公开了一种用于电池模型参数与剩余电量联合异步在线估计方法中的状态方程和观测方程。 [0020] According to a second aspect of the present invention, discloses a battery remaining amount and the model parameters used in a combination of asynchronous state equation and an observation equation of the line estimation method.

[0021] 根据本发明的第三方面,公开了一种用于电池模型参数与剩余电量联合异步在线估计方法所依赖的初始值。 [0021] According to a third aspect of the present invention, there is disclosed an initial value of a battery remaining amount and the model parameters for the asynchronous line Joint estimation depends. 包括电池剩余电量估计的初始化值及电池模型参数估计的初始值等。 It includes a battery remaining amount and the estimated value of the initialization of the battery model parameter estimation initial value. 这些初始值不必很准确,在采样点卡尔曼滤波的后续迭代过程中它们会很快收敛到真实值附近。 These initial values ​​are not necessarily very accurate, in a subsequent iteration of the Kalman filter sampling point they quickly converges to the true value.

[0022] 根据本发明的第四方面,公开了一种应用采样点卡尔曼滤波迭代进行电池模型参数与电池剩余电量联合异步在线估计的具体流程。 [0022] According to a fourth aspect of the invention, the application discloses a sampling point Kalman filter iteration battery remaining battery capacity model parameters specific asynchronous processes combined online estimation. 电池剩余电量的估计依赖于上一时刻电池模型参数的估计结果,而电池模型参数的估计则基于当前时刻所估计得到的电池剩余电量完成,两种估计流程交替异步进行。 Estimating battery residual quantity is dependent on the time a battery model parameter estimation result estimated based on the battery model parameters obtained battery residual quantity estimated time to complete the current two asynchronous alternating flow estimation.

具体实施方式 detailed description

[0023] 电池模型参数与剩余电量联合异步在线估计方法,具体步骤是: 步骤(1)测量在λ时刻的电池端电圧h和电池供电电流4,λ= 。 [0023] and the remaining amount of the battery model parameters asynchronous line Joint estimation method, the specific steps are: Step (1) measured at the battery terminals and a battery power supply current-pressure h time point [lambda] 4, λ =.

[0024] 步骤(2)用状态方程和观测方程表亍电池的各个时刻的荷电状态依赖关系: [0024] Step (2) dependence of the state of charge of each time by a state equation and an observation equation right foot cell table:

Figure CN102289557AD00101

状态方程: Equation of state:

Figure CN102289557AD00102

观测方程 Observation equation

Figure CN102289557AD00103

其中ζ为电池的荷电状态,即剩余电量A为电池的放电比例系数,反映的是放电速率、温度等因素对电池SOC的影响程度,本发明中只考虑放电速率的影响;a是电池在室温25冗条件下、以1/30倍额定电流的放电速率放电时所能得到的额定总电量,Δί是测量时间间隔,Wfc为处理噪声。 Wherein ζ is the state of charge of the battery, i.e., the remaining amount A is a discharge ratio coefficient cells, reflects the degree of influence discharge rate, temperature and other factors on the battery SOC, the present invention considers only affect the rate of discharge; a is battery at room temperature for 25 redundant conditions, a discharge rate of nominal total amount 30 times the rated current of discharge can be obtained, Δί is the measurement time interval, Wfc is the process noise.

Figure CN102289557AD00104

为电池观测模型的参数,是一个列向量;i?为电池的内阻,1为观测噪声。 Cell parameters of the observation model is a column vector; I is the internal resistance of the battery 1 is the observation noise?.

[0025] 放电比例系数访的确定方法为: The method of determining the [0025] discharge ratio coefficient to visit:

(a)将完全充满电的电池以不同放电速率ς (0<ς SC ,C为电池的额定放电电流)恒流放电F(iV>10)次,计算相应放电速率下的电池总电量β , ι仏NO (A) The fully charged battery at different discharge rates ς (0 <ς SC, C is the nominal battery discharge current) constant current discharge F (iV> 10) times, calculates the total battery discharge rate under the corresponding β, ι Fo NO

[0026] (b)根据最小二乘方法拟合出Qi与间的二次曲线关系,即在最小均方误差准则下求出同时满足β = C +hq+c, a,h,c为最优系数。 [0026] (b) according to the least square fitting method with a quadratic curve the relationship between Qi, i.e., determined at minimum mean square error criterion satisfying β = C + hq + c, a, h, c is the most excellent coefficient.

[0027] (c)在放电电流为时对应的放电比例系数"力: [0027] (c) a discharge current in the discharge ratio coefficient corresponding to the last "power:

Vi = -^rzfrr Vi = - ^ rzfrr

aik +hik+c aik + hik + c

此处,由于放电比例系数与电池老化等无关,因此,最优系数“Ae对于同一类型的电 Ae electric Here, since the discharge ratio coefficient irrespective of aging of the battery, and therefore, the optimal coefficient "for the same type of

池只需确定一次,确定后可作为已知常数直接用于所有同类型电池的剩余电量估计。 Determining the pool only once, after the determination may be directly used as a known constant for all the remaining power of the battery of the same type estimation.

[0028] 步骤(3)执行如下初始化过程: (a)电池剩余电量估计的初始化: [0028] Step (3) to perform the initialization process: (a) initialization of the estimated remaining battery capacity:

起始状态為及其方差界分别为: Its initial state is bounded variance were:

Z0 = 100% , P0 = Var(Z0) = 1(Γ2 Z0 = 100%, P0 = Var (Z0) = 1 (Γ2

处理噪声4的方差矣、观测噪声ι的方差氏分别为: Rw = Vr5,民=ICr2 尺度参数ζ为: 4 men process noise variance, variance of the observation noise ι s are: Rw = Vr5, China = ICr2 scale parameter ζ is:

扩展后的状态向量笃及其协方差犮为: The extended state vector and its covariance Benedict c is:

ο ο

Pa Pa

Z = Z =

O of 均值加权系数Mi Mean O of weighting coefficients Mi

OO OO

0 0K. 0 0K.

•ca 2,...,6和方差加权系数O = 0,1 • ca 2, ..., 6, and a weighting coefficient variance O = 0,1

分别为 Respectively

W, W,

(m) if f (M) if f

% %

1<ί <6 1 <ί <6

(b)电池模型参数估计的初始化: 任意选取初始模型参数巩=h (B) initialization of the battery model parameter estimation: arbitrarily selecting initial model parameters Gong = h

设定的平方根均方差矩阵为Sftl, 选取比例常数A , A>1 ; Square-root covariance matrix are set to Sftl, select the proportional constants A, A> 1;

设定变量4 ; Set the variable 4;

4 ;其中Z6为6x6的单位矩阵; 4; wherein Z6 is a 6x6 unit matrix;

设定加权系数iff» Setting the weighting coefficients IFF »

Λ Λ

-7 -7

ι ι

丄W 2k ¥t Shang W 2k ¥ t

j = l,2,---,12 j = l, 2, ---, 12

2hA ” 2k “ 2F [0029] 步骤(4)采用采样点卡尔曼滤波算法进行循环递推: 2hA "2k" 2F [0029] Step (4) The sample point Kalman filter algorithm Recursive:

在时刻λ = 1,2,3,--·,根据测得的电池端电圧^及电池的供电电流4,按下列步骤迭代进行电池模型参数与剩余电量的联合异步估计: (a)电池剩余电量的估计流程 At time λ = 1,2,3, - ·, according to the measured current supply electrical battery terminal and a battery-pressure ^ 4, the following steps are iterated joint asynchronous battery model parameters estimated remaining charge: (a) a battery the remaining capacity estimation flow

①根据时刻的扩展状态向量5L及其协方差 计算该时刻的所有的采样点序歹丨J龙Li : ① all the sample points of the time sequence is calculated according to the extended state vector and its covariance 5L bad moment J Long Shu Li:

Figure CN102289557AD00121

②根据状态方程进行时间域更新: ② updated according to the time-domain equation of state:

由采样点序列Iil1,根据状态方程计算采样点更新3¾^ = /(K-^h) From the sampling point sequence Iil1, sampling point is calculated according to the state update equation 3¾ ^ = / (K- ^ h)

对采样点更新进行加权,计算状态估计: Weighted sampling point update, state estimation is calculated:

Figure CN102289557AD00122

计算状态估计的方差, Calculating the variance state estimation,

Figure CN102289557AD00123

③根据观测方程完成测量更新: ③ The measurement update complete observation equation:

由采样点更新3¾«及1时刻的参数估计值,根据观测方程计算测量更新 Updated parameter estimates 3¾ «1 and the timing by the sampling point, is calculated from the observation equations measurement update

Figure CN102289557AD00124

对测量更新m进行加权,计算测量估计^ M weighted measurement update, calculated measurement evaluation ^

计算测量估计r的方差 R calculated estimated variance of the measurement

Figure CN102289557AD00125

计算f 与饥的互协方差ρ : F calculating cross-covariance hunger and ρ:

Figure CN102289557AD00126

计算卡尔曼增益 Kalman gain calculation

Figure CN102289557AD00127

计算状态更新 Calculation status update

Figure CN102289557AD00128

计算状态更新4的方差 Variance calculation status update 4

Figure CN102289557AD00129

通过上述流程,所得到的状态更新值4即为当前时刻t所估计得到的电池剩余电量。 Through the above process, the resultant state is the updated value of the current time t 4 the estimated battery residual quantity obtained.

[0030] (b)电池模型参数的估计流程: ①计算模型参数的估计值: [0030] The estimation process (b) of the battery model parameters: ① calculating estimates model parameters:

Figure CN102289557AD001210

计算模型参数的平方根均方差矩阵的估计值 Calculate the square root of the model parameters are estimated value of the covariance matrix

Figure CN102289557AD001211

其中, among them,

Figure CN102289557AD001212

为对应矩阵的对角线元素构成的列向量。 Corresponding column vector of the diagonal elements of the matrix configuration.

[0031] ②计算芮的采样点序列3£仏:β;为6X1列向量,为6X6矩阵,故3¾^为6X 13矩阵。 [0031] ② sequence of sampling points calculated Rui £ Fo. 3: β; column vector of 6X1, 6X6 matrix is, it is 3¾ ^ 6X 13 matrix. [0032] ③按下列各式计算测量更新: [0032] ③ measurement update calculated by the following formulas:

计算采样点的观测序列: 2¾^! = g(ik,hD , 为6 X 13矩阵; Observation sequence calculating sampling points: 2¾ ^ = g (ik, hD, of 6 X 13 matrix;!

计算观测序列w 的估计值Ί- ·βΐ = Σ^'^-ι, _ 为Λ的第. Calculating an estimate of the observed sequence w Ί- · βΐ = Σ ^ '^ - ι, _ for the first of Λ.

'tm^-i a λ. /-ο 妨n 'thkr-i j 'Tm ^ -i a λ. / -ο hinder n' thkr-i j

计算观测序列的平方根均方差矩阵: Calculating the square root of the observed sequence covariance matrix are:

^it = ^r (jV 、_ ?Y^L ) (?1:6,耿-1 _ ?1_ 2¾, ^ It = ^ r (jV, _ Y ^ L?) (1:?? 6, Geng -1 _ 1_ 2¾,

计算协方差矩阵i^wf ; 计算卡尔曼增益ζ; :^=(^/¾)/¾ ; 计算参数更新^ :pk=p;+Ki(yk-d^ ; 计算临时变量L/ = ; Covariance matrix i ^ wf; calculating Kalman gain ζ;: ^ = (^ / ¾) / ¾; calculating parameter update ^: pk = p; + Ki (yk-d ^; calculating a temporary variable L / =;

计算模型参数的平方根均方差矩阵的更新Sit :Syt=choIupdai.S(S-pt,Ui}; 其中F【)表示求矩阵的正交三角分解,并返回得到的上三角矩阵;(f为矩阵的转置操作表示求矩阵(¾/ -”rnHf 的Cholesky 分解。 Square-root calculation of the model parameters are updated covariance matrix Sit: Syt = choIupdai.S (S-pt, Ui}; {wherein F) represented by orthogonal trigonometric decomposition of Matrix, and returns the upper triangular matrix obtained; (F matrix It represents the transpose operation of matrix (¾ / - "rnHf the Cholesky decomposition.

[0033] 通过上述流程,所得到的良即为当前时刻λ-所估计得到的电池模型参数。 [0033] Through the above process, the good is the current time obtained by the battery model λ- parameter estimation obtained.

[0034] 在每一时刻,上述步骤4 (a)、4(b)交替进行,因此,电池剩余电量的估计依赖于上一时刻电池模型参数的估计结果,另一方面,电池模型参数的估计则基于当前时刻所估计得到的电池剩余电量完成。 [0034] At each time point, the above step 4 (a), 4 (b) alternately, and therefore, the estimated remaining battery power is estimated dependency estimation result of battery model parameters on one time, on the other hand, the battery model parameters remaining battery power based on the estimated current time is completed. 整个循环递推过程是在线完成的,即在电池实际工作过程中在线异步完成各时刻电池剩余电量的估计与电池模型参数的估计。 Recursive whole process is done online, i.e. during the actual cell line asynchronously estimation and model parameters each time the battery remaining battery capacity.

Claims (1)

  1. 1. 一种电池模型参数与剩余电量联合异步在线估计方法,其特征在于该方法的具体步骤是:步骤(1)测量在 时刻的电池端电压Λ和电池供电电流1免=1义3,…;步骤(¾用状态方程和观测方程表示电池的各个时刻的荷电状态依赖关系: A battery remaining power of the model parameters and the asynchronous line Joint estimation method, characterized in that the specific steps of the method are: Step (1) Λ measured at the battery terminal voltage and the battery current time = 1 1 Free sense 3, ... ; step (¾ a state equation and the observation equation represents the state of charge of the battery at each time dependence:
    Figure CN102289557AC00021
    其中2为电池的荷电状态,即剩余电量”为电池的放电比例系数,反映的是放电速率、温度等因素对电池soc的影响程度,本发明中只考虑放电速率的影响;a是电池在室温25 "C条件下、以1/30倍额定电流的放电速率放电时所能得到的额定总电量· Δ是测量时间间隔,W为处理噪声;R K1 K2 Kj 为电池观测模型的参数,是一个列向量;i?为电池的内阻,Vft为观测噪声;放电比例系数”的确定方法为:(a)将完全充满电的电池以不同放电速率Ci恒流放电Ar次,计算相应放电速率下的电池总电量β ,\<ι<Ν,^ <Ci<C , Λγ>10 , C为电池的额定放电电流;(b)根据最小二乘方法拟合出a与ς间的二次曲线关系,即在最小均方误差准则下求出同时满足fiiC^+M^+c, αAc为最优系数;(c)在放电电流为k时,对应的放电比例系数私为: Wherein 2 is the battery state of charge, i.e. residual amount "for the discharge ratio coefficient cells, reflects the discharge rate, temperature and other factors impact on the battery soc, the present invention considers only affect the rate of discharge; a is battery room temperature of 25 "C under conditions of a discharge rate of 30 times the rated current that can be obtained during discharge · Δ is the nominal total power measurement interval, W is processed noise; R K1 K2 Kj is the observation model parameters of the battery, is a column vector; I is the internal resistance of the battery, Vft for the measurement noise; discharge ratio coefficient determination method "is:? (a) the fully charged battery at different discharge rates Ci Ar constant current discharge times, calculating the corresponding discharge rate β under the total battery charge, \ <ι <Ν, ^ <Ci <C, Λγ> 10, C is the rated discharge current of the battery; (b) fitting a quadratic curve between the least squares method and ς relationship, i.e., at a determined minimum mean square error criterion satisfied fiiC ^ + M ^ + c, αAc optimal coefficient; (c) the discharge current is k, a corresponding private discharge ratio coefficient is:
    Figure CN102289557AC00022
    此处,由于放电比例系数与电池老化无关,因此最优系数《Ac对于同一类型的电池只需确定一次,确定后可作为已知常数直接用于所有同类型电池的剩余电量估计; 步骤(¾执行如下初始化过程: (a)电池剩余电量估计的初始化:起始状态為及其方差药分别为:Z0 = 100%,P0 = var(z0) = IO"2处理噪声>%的方差民.、观测噪声^的方差民分别为: Here, since the discharge ratio coefficient independent of aging of the battery, so the optimum coefficient "Ac determined only once for the same type of battery, the determination may be directly used as a known constant for all the same type of battery remaining amount estimation; step (¾ initialization process performs the following: (a) the estimated remaining battery power initialization: the initial state of the drug and the variance are: Z0 = 100%, P0 = var (z0) = IO "2 noise processing>% variance China. ^ observation noise variance of the people are:
    Figure CN102289557AC00031
    尺度参数 Scale parameter
    Figure CN102289557AC00032
    扩展后的状态向量驾及其协方差疗为: The extended state vector and covariance riding therapy are:
    Figure CN102289557AC00033
    均值加权系数= 0丄2,...,6和方差加权系数=0,1,2,...,6分别为: Mean = 0 Shang weighting factor 2, ..., 6, and a weighting coefficient of variance = 0,1,2, ..., 6 are:
    Figure CN102289557AC00034
    (b)电池模型参数估计的初始化: 任意选取初始模型参数= Λ ;设定i%的平方根均方差矩阵为 (B) initialization of the battery model parameter estimation: arbitrarily selecting initial model parameter = Λ; set i% of the square root of the variance matrix are
    Figure CN102289557AC00035
    其中Z6为6x6的单位矩阵; 选取比例常数λ , k>\ ; 设定变量 Wherein Z6 is a unit matrix of 6x6; Select proportionality constant λ, k> \; set variable
    Figure CN102289557AC00036
    步骤(4)采用采样点卡尔曼滤波算法进行循环递推:在时刻λ = 1,2,\--·,根据测得的电池端电压Λ及电池的供电电流,按下列步骤迭代进行电池模型参数与剩余电量的联合异步估计: (a)电池剩余电量的估计流程①根据时刻的扩展状态向量及其协方差,计算该时刻的所有的采样点序列支“: Step (4) The sample point Kalman filter algorithm Recursive: at the time of λ = 1,2, \ - * the battery model, according to the measured supply current battery terminal voltage and the battery Λ, the following iterative steps combined with the remaining power of the asynchronous parameter estimation: estimation process (a) according to the remaining battery level ① expanded state vector and covariance of the time, all the sample points is calculated the time sequence branched ":
    Figure CN102289557AC00037
    ②根据状态方程进行时间域更新:由采样点序列IJU,根据状态方程计算采样点更新 ② updated according to the time-domain equation of state: the sample point sequence IJU, sampling point is calculated according to the equation of state update
    Figure CN102289557AC00038
    对采样点更新进行加权,计算状态估计t;计算状态估计r的方差 Weighted sampling point update, state estimation is calculated T; r is calculated state estimation variance
    Figure CN102289557AC00039
    ③根据观测方程完成测量更新:由采样点更新3¾«及i —1时刻的参数估计值Jw,根据观测方程计算测量更新 The observation equation ③ complete measurement update: Update the parameter estimates Jw 3¾ «i -1 and timing by the sampling point, is calculated from the observation equations measurement update
    Figure CN102289557AC000310
    :
    Figure CN102289557AC00041
    对测量更新m 进行加权,计算测量估计 M updating weighted measurement, calculation measurement evaluation
    Figure CN102289557AC00042
    计算测量估计扩的方差: Calculate the variance measure is estimated to expand:
    Figure CN102289557AC00043
    计算f 1Jsn 的互协方差ρ : F 1Jsn calculating cross-covariance ρ:
    Figure CN102289557AC00044
    计算卡尔曼增益 Kalman gain calculation
    Figure CN102289557AC00045
    ; 计算状态更新 ; Update calculation state
    Figure CN102289557AC00046
    计算状态更新4的方差 Variance calculation status update 4
    Figure CN102289557AC00047
    通过上述流程,所得到的状态更新值之即为当前时刻fc所估计得到的电池剩余电量; (b)电池模型参数的估计流程:①计算模型参数的估计值λ— ;计算模型参数的平方根均方差矩阵的估计值: S,;= Sfm+Am ,其中, D^ =~diag{SfJ + ^jdmgiSpJ2+djag{,%} , diag{}为对应矩阵的对角线元素构成的列向量;②计算芮的采样点序: Through the above updating process status value, obtained is the current time of the battery residual quantity fc obtained estimation; estimating process (b) the battery model parameters: ① the calculation of the model parameter estimation lambda-value; the square root of the model parameters are the estimated value of the covariance matrix: S,; = Sfm + Am, where, D ^ = ~ diag {SfJ + ^ jdmgiSpJ2 + djag {,%}, diag {} is the matrix corresponding to the column vector composed of diagonal elements; ② sequence of sampling point calculated Rui:
    Figure CN102289557AC00048
    P;为6 X 1列向量,为6 X 6矩阵,故3¾^为6 X 13矩阵;③按下列各式计算测量更新:计算采样点的观测序列: = g(ik,、h) , ^«为6 X 13矩阵; 计算观测序列的估计值:《=为的第.列; P; 6 X 1 is a column vector as a 6 X 6 matrix, it is 3¾ ^ 6 X 13 matrix; ③ measurement update calculated by the following formulas: Calculation observing the sequence of sample points: = g (ik ,, h), ^ «is 6 X 13 matrix; observation sequence calculated estimate:." = is the first column;
    Figure CN102289557AC00049
    计算观测序列^iiw的平方根均方差矩阵:^it = ^r [^/¾、? Calculating the square root of the observation sequence ^ iiw average covariance matrix: ^ it = ^ r [^ / ¾ ,? Imijw ~ ? Imijw ~? j^pi )(迎6,耿-1 一?1¾淋-i ~ 2¾,耿_1)}计算协方差矩阵: ρ?Λ=-效r : 计算卡尔曼增益ζ; =^=(^/¾)/¾ ; 计算参数更新A j ^ pi) (6 Ying, Geng -1 1¾ a shower -i ~ 2¾, Geng _1)} covariance matrix is ​​calculated: ρ Λ = - Validity r: calculating Kalman gain ζ; = ^ = (^ /?? ¾) / ¾; update calculation parameters A
    Figure CN102289557AC000410
    计算临时变量了:U =Ks3t ;计算模型参数的平方根均方差矩阵的更新 Calculate temporary variables: U = Ks3t; calculate the square root of the model parameters are updated covariance matrix
    Figure CN102289557AC00051
    其中FO表示求矩阵的正交三角分解,并返回得到的上三角矩阵;Cf为矩阵的转置操作表示求矩阵 Wherein the FO denotes orthogonal trigonometric decomposition of Matrix, and returns the upper triangular matrix obtained; of Cf is represented by a matrix transpose operation of Matrix
    Figure CN102289557AC00052
    的分解;通过上述流程,所得到的即为当前时刻> 所估计得到的电池模型参数;在每一时刻,上述步骤4(a) ,4(b)交替进行,因此,电池剩余电量的估计依赖于上一时刻电池模型参数的估计结果,另一方面,电池模型参数的估计则基于当前时刻所估计得到的电池剩余电量完成;整个循环递推过程是在线完成的,即在电池实际工作过程中在线异步完成各时刻电池剩余电量的估计与电池模型参数的估计。 Decomposition; by the above process, the resultant is the current time> obtained by the battery model parameter estimation; at each moment, the above step 4 (a), 4 (b) alternately, therefore, dependent on the estimated remaining battery power on a time model parameters estimation result of the battery, on the other hand, the estimated battery current model parameters based on the estimated time to complete the remaining battery power obtained; recursive whole process is done online, i.e. during the actual cell battery model parameter estimation and online asynchronous completion of each moment of remaining battery power.
CN 201110127479 2011-05-17 2011-05-17 Battery model parameter and residual battery capacity joint asynchronous online estimation method CN102289557B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201110127479 CN102289557B (en) 2011-05-17 2011-05-17 Battery model parameter and residual battery capacity joint asynchronous online estimation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201110127479 CN102289557B (en) 2011-05-17 2011-05-17 Battery model parameter and residual battery capacity joint asynchronous online estimation method

Publications (2)

Publication Number Publication Date
CN102289557A true true CN102289557A (en) 2011-12-21
CN102289557B CN102289557B (en) 2013-08-07

Family

ID=45335981

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201110127479 CN102289557B (en) 2011-05-17 2011-05-17 Battery model parameter and residual battery capacity joint asynchronous online estimation method

Country Status (1)

Country Link
CN (1) CN102289557B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102298118A (en) * 2011-05-17 2011-12-28 杭州电子科技大学 A battery model parameters and the remaining capacity estimation method with concurrent online
CN102928783A (en) * 2012-07-19 2013-02-13 北京金山安全软件有限公司 Method and device for estimating usable time of battery power and mobile equipment comprising device
CN103077291A (en) * 2013-01-25 2013-05-01 华北电力大学 Battery charge and discharge process digital simulation method capable of setting initial state of charge
WO2015180050A1 (en) * 2014-05-26 2015-12-03 北京理工大学 Method for estimating parameters and state of dynamical system of electric vehicle

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000075967A (en) * 1998-08-27 2000-03-14 Matsushita Electric Ind Co Ltd Method and device for evaluating battery residual capacity of portable personal computer
CN101022178A (en) * 2007-03-09 2007-08-22 清华大学 Method for estimating nickel-hydrogen power battery charged state based on standard battery model
CN101212071A (en) * 2006-12-31 2008-07-02 比亚迪股份有限公司 Method for estimating charge state of power cell
CN101598769A (en) * 2009-06-29 2009-12-09 杭州电子科技大学 Method for estimating remaining capacity of battery based on sampling points Kalman filtering
CN101604005A (en) * 2009-06-29 2009-12-16 杭州电子科技大学 Estimation method of battery dump energy based on combined sampling point Kalman filtering
CN101625397A (en) * 2009-08-06 2010-01-13 杭州电子科技大学 Mixed rapid estimation method for residual energy of battery
JP2010048759A (en) * 2008-08-25 2010-03-04 Nippon Telegr & Teleph Corp <Ntt> Residual capacity estimation technique and residual capacity estimating device

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000075967A (en) * 1998-08-27 2000-03-14 Matsushita Electric Ind Co Ltd Method and device for evaluating battery residual capacity of portable personal computer
CN101212071A (en) * 2006-12-31 2008-07-02 比亚迪股份有限公司 Method for estimating charge state of power cell
CN101022178A (en) * 2007-03-09 2007-08-22 清华大学 Method for estimating nickel-hydrogen power battery charged state based on standard battery model
JP2010048759A (en) * 2008-08-25 2010-03-04 Nippon Telegr & Teleph Corp <Ntt> Residual capacity estimation technique and residual capacity estimating device
CN101598769A (en) * 2009-06-29 2009-12-09 杭州电子科技大学 Method for estimating remaining capacity of battery based on sampling points Kalman filtering
CN101604005A (en) * 2009-06-29 2009-12-16 杭州电子科技大学 Estimation method of battery dump energy based on combined sampling point Kalman filtering
CN101625397A (en) * 2009-08-06 2010-01-13 杭州电子科技大学 Mixed rapid estimation method for residual energy of battery

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
《浙江省电源学会第十一届学术年会暨省科协重点科技活动"高效节能电力电子新技术"研讨会论文集》 20081231 程艳青 等 基于卡尔曼滤波的电动汽车SOC估计 24-27 1 , *
刘建锋 等: "基于异步估计自适应Kalman滤波系统设计", 《系统工程与电子技术》, vol. 30, no. 10, 31 October 2008 (2008-10-31), pages 1824 - 1827 *
程艳青 等: "基于卡尔曼滤波的电动汽车SOC估计", 《浙江省电源学会第十一届学术年会暨省科协重点科技活动"高效节能电力电子新技术"研讨会论文集》, 31 December 2008 (2008-12-31), pages 24 - 27 *
程艳青 等: "基于卡尔曼滤波的电动汽车剩余电量估计", 《杭州电子科技大学学报》, vol. 29, no. 3, 30 June 2009 (2009-06-30), pages 4 - 7 *
程艳青 等: "电动汽车动力电池剩余电量在线测量", 《电子测量与仪器学报》, vol. 2008, no. 2, 31 December 2008 (2008-12-31), pages 182 - 185 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102298118A (en) * 2011-05-17 2011-12-28 杭州电子科技大学 A battery model parameters and the remaining capacity estimation method with concurrent online
CN102928783A (en) * 2012-07-19 2013-02-13 北京金山安全软件有限公司 Method and device for estimating usable time of battery power and mobile equipment comprising device
CN102928783B (en) 2012-07-19 2014-09-03 北京金山安全软件有限公司 Method and device for estimating usable time of battery power and mobile equipment comprising device
CN103077291A (en) * 2013-01-25 2013-05-01 华北电力大学 Battery charge and discharge process digital simulation method capable of setting initial state of charge
CN103077291B (en) * 2013-01-25 2016-05-18 华北电力大学 You can set the initial state of charge of the battery during Simulation Method
WO2015180050A1 (en) * 2014-05-26 2015-12-03 北京理工大学 Method for estimating parameters and state of dynamical system of electric vehicle

Also Published As

Publication number Publication date Type
CN102289557B (en) 2013-08-07 grant

Similar Documents

Publication Publication Date Title
He et al. A new model for State-of-Charge (SOC) estimation for high-power Li-ion batteries
Kinoshita et al. Eighth-order QED contribution to the anomalous magnetic moment of the muon
Marcicki et al. Design and parametrization analysis of a reduced-order electrochemical model of graphite/LiFePO4 cells for SOC/SOH estimation
CN102445663A (en) Method for estimating battery health of electric automobile
JPH11346444A (en) Estimating method of battery charged condition
CN102930173A (en) SOC(state of charge) online estimation method for lithium ion battery
CN101383511A (en) Electrical power system status estimation method based on measurement data by data collection system
CN102520361A (en) State of health (SOH) value assessment method of battery pack
CN101975927A (en) Method and system for estimating remaining available capacity of lithium ion power battery pack
CN102121973A (en) Method for testing net energy of electrical vehicle power battery
CN103529398A (en) Online lithium ion battery SOC (state of charge) estimation method based on extended Kalman filter
CN101965522A (en) Method for Measuring SOC of a Battery in a Battery Management System and the Apparatus Thereof
CN103163480A (en) Method for estimating health state of lithium battery
CN103197257A (en) Method and device for detecting state of health (SOH) of battery
CN103744026A (en) Storage battery state of charge estimation method based on self-adaptive unscented Kalman filtering
CN102645637A (en) Method for estimating SOC (state of charge) of equalized batteries
CN102662148A (en) On-line feedback battery state of charge (SOC) predicting method
CN102540096A (en) Self-correction method for remaining capacity estimation of lithium iron phosphate power battery
JP2010203854A (en) Device of estimating internal state of secondary battery
CN103472398A (en) Power battery SOC (state of charge) estimation method based on expansion Kalman particle filter algorithm
CN102135603A (en) Device for estimating cycle life of battery
CN102540083A (en) Battery management system and method of estimating battery state of charge
CN102937704A (en) Method for identifying RC (resistor-capacitor) equivalent model of power battery
CN102916427A (en) Method for calculating integrated three-phase imbalance degree of multi-circuit transmission lines
CN103353918A (en) Method for selecting initial values of three-phase Newton-Raphson method power flow calculation of power distribution network

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model