CN105891715A

CN105891715A - Lithium ion battery health state estimation method

Info

Publication number: CN105891715A
Application number: CN201410758219.8A
Authority: CN
Inventors: 潘海鸿; 林伟龙; 李海波; 陈琳; 李君子; 黄炳琼
Original assignee: Guangxi University
Current assignee: Guangxi University
Priority date: 2014-12-12
Filing date: 2014-12-12
Publication date: 2016-08-24

Abstract

The invention proposes a method for estimating the state of health of a lithium-ion battery, the main steps of which are: (1) establishing an equivalent physical model of the battery; (2) obtaining the battery SOC-OCV curve through battery pulse discharge, and obtaining the battery SOC and open circuit Voltage relationship; (3) Collect battery voltage, current and temperature parameters; (4) Based on the recursive least squares method, identify model parameters online to obtain battery ohmic internal resistance; (5) Use the estimated value of battery ohmic internal resistance to calculate battery health status. The invention can estimate the health state of the battery on-line, and when the method estimates the health state of the battery, the working conditions used are random, and the estimation result can accurately reflect the real health state of the battery.

Description

A method for estimating the state of health of lithium-ion batteries

技术领域technical field

本发明属于新能源电池管理系统领域，涉及一种电池健康状态的在线估算方法。The invention belongs to the field of new energy battery management systems, and relates to an online estimation method of battery health status.

背景技术Background technique

源危机与环境污染问题为新能源汽车发展带来了契机，锂离子电池以其工作电压高、质量轻、体积小、比能量大、寿命长等优势成为新能源汽车动力源的不二选择，然而，由于电池管理系统中对电池健康状态监测不力引发的安全事故屡见不鲜，如何有效估算电池健康状态，对锂离子电池在新能源汽车应用中推广具有重要意义。The source crisis and environmental pollution problems have brought opportunities for the development of new energy vehicles. Lithium-ion batteries have become the best choice for new energy vehicles due to their advantages such as high working voltage, light weight, small size, large specific energy, and long life. However, due to the frequent safety accidents caused by poor monitoring of the battery health status in the battery management system, how to effectively estimate the battery health status is of great significance to the promotion of lithium-ion batteries in new energy vehicle applications.

电池健康状态表征了电池老化程度，从工作原理分析，锂离子电池老化的主要原因为：电池正极金属阳离子与内部电解质作用，产生副反应效应并溶解于电解质中，且在电池循环工作过程中，金属阳离子与电池负极发生氧化还原反应，在电解质表面形成界面膜(SEI)，减少了电池内部活性锂离子数量。从电池使用条件分析，加速电池老化的原因有：高温或低温环境；过充与过放；高倍率充放电。电池老化是一个缓慢的、不可逆变化过程，随着电池组循环使用，电池单体健康程度差异将逐步加大，单体差异性增大使电池组使用效率减小、寿命缩减，构成恶性循环。就目前的研究成果看，电池健康状态在线估算研究较少，大部分研究都利用实验数据离线分析电池健康状态，无法实现电池健康状态在电池管理系统中的应用及推广。针对此种情况，建立电池组模型，实现电池组健康状态的在线估算已然成为动力电池领域研究人员关注的焦点，对电动汽车发展具有重要意义。The health status of the battery represents the aging degree of the battery. From the analysis of the working principle, the main reason for the aging of the lithium-ion battery is: the positive metal cation of the battery interacts with the internal electrolyte, which produces a side reaction effect and dissolves in the electrolyte, and during the battery cycle, The redox reaction between metal cations and the negative electrode of the battery forms an interfacial film (SEI) on the surface of the electrolyte, reducing the number of active lithium ions inside the battery. From the analysis of battery usage conditions, the reasons for accelerated battery aging are: high temperature or low temperature environment; overcharge and overdischarge; high rate charge and discharge. Battery aging is a slow and irreversible change process. With the repeated use of the battery pack, the difference in the health of the battery cells will gradually increase. The increase in the difference between the cells will reduce the efficiency and life of the battery pack, forming a vicious circle. As far as the current research results are concerned, there are few studies on online estimation of battery health status, and most studies use experimental data to analyze battery health status offline, which cannot realize the application and promotion of battery health status in battery management systems. In view of this situation, establishing a battery pack model and realizing online estimation of battery pack health status has become the focus of researchers in the field of power batteries, which is of great significance to the development of electric vehicles.

发明内容Contents of the invention

本发明提出一种锂离子电池健康状态估算方法，其通过建立电池等效物理模型，利用递推最小二乘法在线辨识模型参数，进而估算电池欧姆内阻，结合电池欧姆内阻与电池健康状态关系估算电池健康状态。The invention proposes a method for estimating the state of health of a lithium-ion battery, which establishes an equivalent physical model of the battery, uses the recursive least squares method to identify model parameters online, and then estimates the ohmic internal resistance of the battery, combining the relationship between the ohmic internal resistance of the battery and the healthy state of the battery Estimate battery state of health.

本发明一种锂离子电池健康状态估算方法，其估算方法包括如下步骤：A method for estimating the state of health of a lithium-ion battery of the present invention, the estimating method comprising the steps of:

步骤1：建立电池等效物理模型；Step 1: Establish an equivalent physical model of the battery;

步骤2：通过电池脉冲放电获取当电池OCV-SOC曲线，并求取电池SOC与电池开路电压关系式；Step 2: Obtain the current battery OCV-SOC curve through battery pulse discharge, and obtain the relationship between battery SOC and battery open circuit voltage;

步骤3：采集电池电压、电流和温度参数；Step 3: Collect battery voltage, current and temperature parameters;

步骤4：以递推最小二乘法为基础在线辨识模型参数，获取电池欧姆内阻R_ser；Step 4: Online identification of model parameters based on the recursive least squares method to obtain the ohmic internal resistance R _ser of the battery;

根据电池等效物理模型，采用带遗忘因子的递推最小二乘算法，用OCV表示电池的开路电压，其由U_oc和C_cap两部分电压组成，电流I为模型的输入激励，令y＝OCV-U_Bat，则y表示模型的响应，根据拉普拉斯变化，可得模型的频域表达式(1)：According to the equivalent physical model of the battery, the recursive least squares algorithm with forgetting factor is used, and the open circuit voltage of the battery is represented by OCV, which is composed of two parts of voltage U _oc and C _cap , and the current I is the input excitation of the model. Let y= OCV-U _Bat , then y represents the response of the model. According to the Laplace change, the frequency domain expression (1) of the model can be obtained:

$y the y ((s the s)) = = I I ((s the s)) (({R R}_{ser ser} + + \frac{{R R}_{d d}}{11 + + {R R}_{d d} {C C}_{d d} s the s} + + \frac{{R R}_{e e}}{11 + + {R R}_{e e} {C C}_{e e} s the s})) - - - - - - ((11))$

则模型的传递函数表达式为(2)：Then the transfer function expression of the model is (2):

$G G ((s the s)) = = {R R}_{ser ser} + + \frac{{R R}_{d d}}{11 + + {R R}_{d d} {C C}_{d d} s the s} + + \frac{{R R}_{e e}}{11 + + {R R}_{e e} {C C}_{e e} s the s} - - - - - - ((22))$

取δ_d＝R_dC_d，δ_e＝R_eC_e，将表达式(2)变形可得：Taking δ _d ＝R _d C _d , δ _e ＝R _e C _e , transforming the expression (2) to get:

$G G ((s the s)) = = \frac{{R R}_{ser ser} {s the s}^{22} + + \frac{{R R}_{ser ser} (({δ δ}_{d d} + + {δ δ}_{e e})) + + {R R}_{d d} {δ δ}_{e e} + + {R R}_{e e} {δ δ}_{d d}}{{δ δ}_{d d} {δ δ}_{e e}} s the s + + \frac{{R R}_{ser ser} + + {δ δ}_{d d} + + {δ δ}_{e e}}{{δ δ}_{d d} {δ δ}_{e e}}}{{s the s}^{22} + + \frac{{δ δ}_{d d} + + {δ δ}_{e e}}{{δ δ}_{d d} {δ δ}_{e e}} s the s + + \frac{11}{{δ δ}_{d d} {δ δ}_{e e}}} - - - - - - ((33))$

由双线性变换法工作原理可知，采用双线性变换法对电池等效物理模型处理后得到的离散传递函数和原连续传递函数具有相同阶数，当电池参数采样周期较小时，通过对比，上述两种传递函数较为相似，故可用双线性变换法对表达式(3)作离散化处理。It can be seen from the working principle of the bilinear transformation method that the discrete transfer function obtained after processing the equivalent physical model of the battery with the bilinear transformation method has the same order as the original continuous transfer function. When the battery parameter sampling period is small, by comparison, The above two transfer functions are relatively similar, so the expression (3) can be discretized by the bilinear transformation method.

令对原传递函数离散化处理，可得：make By discretizing the original transfer function, we can get:

$G G (({z z}^{- - 11})) = = \frac{{γ γ}_{00} + + {γ γ}_{11} {z z}^{- - 11} + + {γ γ}_{22} {z z}^{- - 22}}{11 + + {λ λ}_{11} {z z}^{- - 11} + + {λ λ}_{22} {z z}^{- - 22}} - - - - - - ((44))$

其中，λ₁，λ₂，γ₀，γ₁，γ₂均为待定参数。Among them, λ ₁ , λ ₂ , γ ₀ , γ ₁ , and γ ₂ are all undetermined parameters.

由表达式(4)可得对频域表达式离散化处理后的差分方程为(5)：From the expression (4), it can be obtained that the difference equation after the discretization of the frequency domain expression is (5):

y(k)＝-λ₁y(k-1)-λ₂y(k-2)+γ₀I(k)+γ₁I(k-1)+γ₂I(k-2) (5)y(k)＝-λ ₁ y(k-1)-λ ₂ y(k-2)+γ ₀ I(k)+γ ₁ I(k-1)+γ ₂ I(k-2) (5 )

令make

$\{\begin{matrix} θ θ = = [\begin{matrix} {λ λ}_{11} & {λ λ}_{22} & {γ γ}_{00} & {γ γ}_{11} & {γ γ}_{22} \end{matrix}] \\ h h ((k k)) = = [\begin{matrix} y the y ((k k - - 11)) & y the y ((k k - - 22)) & I I ((k k)) & I I ((k k - - 11)) & I I ((k k - - 22)) \end{matrix}] \end{matrix} - - - - - - ((66))$

假设在k时刻电池参数测试的采样结果误差为e(k)，并将表达式(5)写成最小二乘格式y(k)＝h^T(k)θ+e(k)，利用带遗忘因子的递推最小二乘法可得参数λ₁，λ₂，γ₀，γ₁，γ₂值；Assuming that the error of the sampling result of the battery parameter test at time k is e(k), and the expression (5) is written in the least square format y(k)=h ^T (k)θ+e(k), using the forgetting factor The recursive least square method can get the parameters λ ₁ , λ ₂ , γ ₀ , γ ₁ , γ ₂ values;

令代入表达式(4)，双线性逆变换，可得：make Substituting expression (4) and bilinear inverse transformation, we can get:

$G G ((s the s)) = = \frac{{T T}^{22} (({γ γ}_{00} - - {γ γ}_{11} + + {γ γ}_{22})) {s the s}^{22} + + 44 T T (({γ γ}_{00} - - {γ γ}_{22})) + + 44 (({γ γ}_{00} + + {γ γ}_{11} + + {γ γ}_{22}))}{{T T}^{22} ((11 - - {λ λ}_{11} + + {λ λ}_{22})) {s the s}^{22} + + 44 T T ((11 - - {λ λ}_{22})) s the s + + 44 ((11 + + {λ λ}_{11} + + {λ λ}_{22}))} - - - - - - ((77))$

将该表达式(7)以表达式(3)的形式化简处理，得：Simplifying the expression (7) in the form of expression (3), we get:

$G G ((s the s)) = = \frac{\frac{(({γ γ}_{00} - - {γ γ}_{11} + + {γ γ}_{22}))}{11 - - {λ λ}_{11} + + {λ λ}_{22}} {s the s}^{22} + + \frac{44 (({γ γ}_{00} - - {γ γ}_{22}))}{T T ((11 - - {λ λ}_{11} + + {λ λ}_{22}))} s the s + + \frac{44 (({γ γ}_{00} + + {γ γ}_{11} + + {γ γ}_{22}))}{{T T}^{22} ((11 - - {λ λ}_{11} + + {λ λ}_{22}))}}{{s the s}^{22} + + \frac{44 ((11 - - {λ λ}_{22}))}{T T ((11 - - {λ λ}_{11} + + {λ λ}_{22}))} s the s + + \frac{44 ((11 + + {λ λ}_{11} + + {λ λ}_{22}))}{{T T}^{22} ((11 - - {λ λ}_{11} + + {λ λ}_{22}))}} - - - - - - ((88))$

对比表达式(3)和(8)，可得：Comparing expressions (3) and (8), we can get:

$\{\begin{matrix} {R R}_{ser ser} = = \frac{{γ γ}_{00} - - {γ γ}_{11} + + {γ γ}_{22}}{11 - - {λ λ}_{11} + + {λ λ}_{22}} \\ {δ δ}_{d d} {δ δ}_{e e} = = \frac{{T T}^{22} ((11 - - {λ λ}_{11} + + {λ λ}_{22}))}{44 ((11 + + {λ λ}_{11} + + {λ λ}_{22}))} \\ {δ δ}_{d d} + + {δ δ}_{e e} = = \frac{T T ((11 - - {λ λ}_{22}))}{((11 + + {λ λ}_{11} + + {λ λ}_{22}))} \\ {R R}_{ser ser} + + {R R}_{d d} + + {R R}_{e e} = = \frac{{γ γ}_{00} + + {γ γ}_{11} + + {γ γ}_{22}}{11 + + {λ λ}_{11} + + {λ λ}_{22}} \\ {R R}_{ser ser} (({δ δ}_{d d} + + {δ δ}_{e e})) + + {R R}_{d d} {δ δ}_{e e} + + {R R}_{e e} {δ δ}_{d d} = = \frac{T T (({γ γ}_{00} - - {γ γ}_{22}))}{11 + + {λ λ}_{11} + + {λ λ}_{22}} \end{matrix} - - - - - - ((99))$

其中，T为实验采样周期，由表达式(9)可求得R_ser，R_d，R_e，δ_d，δ_e的值，再根据δ_d＝R_dC_d，δ_e＝R_eC_e，即可得模型参数R_ser，R_d，R_e，C_d，C_e。Among them, T is the experimental sampling period, the values of R _ser , R _d , _Re , δ _d , δ _e can be obtained from the expression (9), and then according to δ _d =R _d C _d , δ _e =R _e C _e , the model parameters R _ser , R _d , R _e , C _d , and C _e can be obtained.

步骤5：利用电池欧姆内阻估算值计算电池健康状态；Step 5: Calculate the state of health of the battery using the estimated ohmic internal resistance of the battery;

$SOH SOH = = \frac{{R R}_{old old} - - {R R}_{cur cur}}{{R R}_{old old} - - {R R}_{new new}} \times \times 100100 % % - - - - - - ((1010))$

R_cur代表当前电池欧姆内阻，R_new表示电池出厂时的欧姆内阻，R_old表示电池容量下降到80％时的欧姆内阻。R _cur represents the ohmic internal resistance of the current battery, R _new represents the ohmic internal resistance of the battery when it leaves the factory, and R _old represents the ohmic internal resistance when the battery capacity drops to 80%.

本发明的有益效果是：The beneficial effects of the present invention are:

1.本发明一种锂离子电池健康状态估算方法实现了锂离子电池健康状态在线估算，使电池管理系统实时监测电池健康状态成为可能；1. A method for estimating the state of health of a lithium-ion battery according to the present invention realizes online estimation of the state of health of the lithium-ion battery, making it possible for the battery management system to monitor the state of health of the battery in real time;

2.本发明一种锂离子电池健康状态估算方法，该方法估算电池健康状态时，所用工况具有随机性，使电池健康状态估算不再局限于某一特定工况。2. A method for estimating the state of health of a lithium-ion battery according to the present invention. When estimating the state of health of the battery, the working conditions used are random, so that the estimation of the state of health of the battery is no longer limited to a specific working condition.

附图说明Description of drawings

图1为锂离子电池健康状态估算方法流程图Figure 1 is a flow chart of the method for estimating the state of health of lithium-ion batteries

具体实施例specific embodiment

为便于本领域技术人员的理解，下面结合附图1和具体实施例，描述本发明一种锂离子电池健康状态估算方法。其估算方法包括如下步骤：In order to facilitate the understanding of those skilled in the art, a method for estimating the state of health of a lithium-ion battery according to the present invention will be described below in conjunction with FIG. 1 and specific embodiments. Its estimation method includes the following steps:

步骤3：采集电池电压、电流和温度参数，采样频率可以为0.01Hz到10KHz，具体数值以实际需求而定，本实施例采样频率选取为2Hz；Step 3: Collect the battery voltage, current and temperature parameters. The sampling frequency can be 0.01Hz to 10KHz. The specific value depends on the actual demand. The sampling frequency in this embodiment is selected as 2Hz;

令make

在此说明书中，应当指出，以上实施例仅是本发明较有代表性的例子。显然本发明不局限于上述具体实施例，还可以做出各种修改、变换和变形。因此，说明书和附图应该被认为是说明性的而非限制性的。凡是依据本发明的技术实质对以上实施例所作的任何简单修改、等同变化与修饰，均应认为属于本发明的保护范围。In this specification, it should be pointed out that the above embodiments are only representative examples of the present invention. It is obvious that the present invention is not limited to the above specific embodiments, and various modifications, changes and variations can be made. Accordingly, the specification and drawings are to be regarded as illustrative rather than restrictive. Any simple modifications, equivalent changes and modifications made to the above embodiments according to the technical essence of the present invention shall be deemed to belong to the protection scope of the present invention.

Claims

1. a health state of lithium ion battery evaluation method, by setting up battery equivalent physical model, estimates battery ohmic internal resistance, In conjunction with battery ohmic internal resistance and cell health state relation estimating state of health of battery, it is characterised in that described estimation battery ohm Internal resistance method is to utilize least square method of recursion on-line identification model parameter, and then estimation health state of lithium ion battery.

Estimation health state of lithium ion battery the most according to claim 1, it is characterised in that its evaluation method includes as follows Step:

Step 1: set up battery equivalent physical model；

Step 2: obtained when battery OCV-SOC curve by cell pulse discharge, and ask for battery SOC and battery open circuit electricity Pressure relational expression；

Step 3: gather cell voltage, electric current and temperature parameter；

Step 4: on-line identification model parameter based on least square method of recursion, obtains battery ohmic internal resistance R_ser；

According to battery equivalent physical model, use the RLS of band forgetting factor, represent the open circuit of battery with OCV Voltage, it is by U_ocAnd C_capTwo parts voltage forms, and electric current I is the input stimulus of model, makes y=OCV-U_Bat, then y table The response of representation model, according to laplace transform, can obtain the frequency-domain expression (1) of model:

y (s) = I (s) (R_{ser} + \frac{R_{d}}{1 + R_{d} C_{d} s} + \frac{R_{e}}{1 + R_{e} C_{e} s}) - - - (1)

Then the transmission function expression of model is (2):

G (s) = R_{ser} + \frac{R_{d}}{1 + R_{d} C_{d} s} + \frac{R_{e}}{1 + R_{e} C_{e} s} - - - (2)

Take δ_d=R_dC_d, δ_e=R_eC_e, expression formula (2) deformation can be obtained:

G (s) = \frac{R_{ser} s^{2} + \frac{R_{ser} (δ_{d} + δ_{e}) + R_{d} δ_{e} + R_{e} δ_{d}}{δ_{d} δ_{e}} s + \frac{R_{ser} + δ_{d} + δ_{e}}{δ_{d} δ_{e}}}{s^{2} + \frac{δ_{d} + δ_{e}}{δ_{d} δ_{e}} s + \frac{1}{δ_{d} δ_{e}}} - - - (3)

From Bilinear transformation method operation principle, use that Bilinear transformation method obtains after processing battery equivalent physical model from Dissipate transmission function and former continuous transmission function has identical exponent number, when the battery parameter sampling period is less, by contrast, above-mentioned Two kinds of transmission functions are the most similar, therefore expression formula (3) is made sliding-model control by available Bilinear transformation method.

OrderTo former transmission function sliding-model control, can obtain:

G (z^{- 1}) = \frac{γ_{0} + γ_{1} z^{- 1} + γ_{2} z^{- 2}}{1 + λ_{1} z^{- 1} + λ_{2} z^{- 2}} - - - (4)

Wherein, λ₁, λ₂, γ₀, γ₁, γ₂It is undetermined parameter.

Can be (5) to the difference equation after frequency-domain expression sliding-model control by expression formula (4):

Y (k)=-λ₁y(k-1)-λ₂y(k-2)+γ₀I(k)+γ₁I(k-1)+γ₂I(k-2) (5)

Order

\{\begin{matrix} θ = [\begin{matrix} λ_{1} & λ_{2} & γ_{0} & γ_{1} & γ_{2} \end{matrix}] \\ h (k) = [\begin{matrix} y (k - 1) & y (k - 2) & I (k) & I (k - 1) & I (k - 2) \end{matrix}] \end{matrix} - - - (6)

Assume that the sampled result error in the test of k moment battery parameter is e (k), and write expression formula (5) as least square form Y (k)=h^TK () θ+e (k), utilizes the least square method of recursion of band forgetting factor can obtain parameter lambda₁, λ₂, γ₀, γ₁, γ₂Value；

OrderSubstitute into expression formula (4), bilinearity inverse transformation, can obtain:

G (s) = \frac{T^{2} (γ_{0} - γ_{1} + γ_{2}) s^{2} + 4 T (γ_{0} - γ_{2}) + 4 (γ_{0} + γ_{1} + γ_{2})}{T^{2} (1 - λ_{1} + λ_{2}) s^{2} + 4 T (1 - λ_{2}) s + 4 (1 + λ_{1} + λ_{2})} - - - (7)

This expression formula (7) is processed with the form abbreviation of expression formula (3):

G (s) = \frac{\frac{(γ_{0} - γ_{1} + γ_{2})}{1 - λ_{1} + λ_{2}} s^{2} + \frac{4 (γ_{0} - γ_{2})}{T (1 - λ_{1} + λ_{2})} s + \frac{4 (γ_{0} + γ_{1} + γ_{2})}{T^{2} (1 - λ_{1} + λ_{2})}}{s^{2} + \frac{4 (1 - λ_{2})}{T (1 - λ_{1} + λ_{2})} s + \frac{4 (1 + λ_{1} + λ_{2})}{T^{2} (1 - λ_{1} + λ_{2})}} - - - (8)

Contrast expression formula (3) and (8), can obtain:

\{\begin{matrix} R_{ser} = \frac{γ_{0} - γ_{1} + γ_{2}}{1 - λ_{1} + λ_{2}} \\ δ_{d} δ_{e} = \frac{T^{2} (1 - λ_{1} + λ_{2})}{4 (1 + λ_{1} + λ_{2})} \\ δ_{d} + δ_{e} = \frac{T (1 - λ_{2})}{(1 + λ_{1} + λ_{2})} \\ R_{ser} + R_{d} + R_{e} = \frac{γ_{0} + γ_{1} + γ_{2}}{1 + λ_{1} + λ_{2}} \\ R_{ser} (δ_{d} + δ_{e}) + R_{d} δ_{e} + R_{e} δ_{d} = \frac{T (γ_{0} - γ_{2})}{1 + λ_{1} + λ_{2}} \end{matrix} - - - (9)

Wherein, T is the experiment sampling period, expression formula (9) can try to achieve R_ser, R_d, R_e, δ_d, δ_eValue, further according to δ_d=R_dC_d, δ_e=R_eC_e, model parameter R_ser, R_d, R_e, C_d, C_e。

Step 5: utilize battery ohmic internal resistance estimated value to calculate cell health state；

SOH = \frac{R_{old} - R_{cur}}{R_{old} - R_{new}} \times 100 % - - - (10)

R_curRepresent present battery ohmic internal resistance, R_newRepresent ohmic internal resistance when battery dispatches from the factory, R_oldRepresent that battery capacity drops to Ohmic internal resistance when 80%.