WO2021259196A1

WO2021259196A1 - Battery pack consistency evaluation method and system

Info

Publication number: WO2021259196A1
Application number: PCT/CN2021/101218
Authority: WO
Inventors: 张雷; 王震坡; 王秋诗
Original assignee: 北京理工大学
Priority date: 2020-06-22
Filing date: 2021-06-21
Publication date: 2021-12-30
Also published as: CN111707951B; CN111707951A

Abstract

A battery pack consistency evaluation method and system. The method comprises: dividing a vehicle state to obtain a fully-charged standing segment, a driving segment, and a constant-current charging segment (S1); extracting an open-circuit voltage of each battery cell by means of a battery cell voltage on the basis of the fully-charged standing segment (S2); on the basis of the driving fragment, establishing an equivalent circuit model for the battery cells, performing parameter identification on the battery cells by using a self-adaptive OCV-RLS method, and extracting ohmic internal resistance of each battery cell (S3); calculating a charging voltage vector norm of each battery cell on the basis of the constant-current charging segment (S4); according to the open-circuit voltage, the ohmic internal resistance, and the charging voltage vector norm of each battery cell, calculating the consistency of all battery packs to be tested at evaluation points in combination with a Mahalanobis distance and a DBSCAN clustering algorithm (S5); calculating the Z scores of the battery packs to be tested at the evaluation points according to the consistency of the battery packs to be tested at the evaluation points (S6); and determining the consistency of the battery packs to be tested according to the Z scores (S7). An abnormal OCV identification value is compensated on the basis of the self-adaptive OCV-RLS method, thereby improving the accuracy of the battery pack consistency evaluation result.

Description

一种电池组一致性评估方法及***Method and system for evaluating battery pack consistency

本申请要求于2020年06月22日提交中国专利局、申请号为202010572859.5、发明名称为“一种电池组一致性评估方法及***”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on June 22, 2020, the application number is 202010572859.5, and the invention title is "a method and system for evaluating battery pack consistency", the entire content of which is incorporated by reference In this application.

技术领域Technical field

本发明涉及电池评估技术领域，特别是涉及一种电池组一致性评估方法及***。The present invention relates to the technical field of battery evaluation, in particular to a battery pack consistency evaluation method and system.

背景技术Background technique

锂离子电池由于其能量密度高、循环寿命长、自放电率低等特点，逐渐成为目前应用最为广泛的电动汽车动力电池类型。为满足电动汽车长续航和高功率的使用要求，车载锂离子动力电池组通常由若干个电池单体串并联而成。由于制造和使用过程的差异，车载动力电池组内大量电池单体不可避免地会存在性能不一致的问题，其具有多方面表征，如容量不一致、内阻不一致、电压不一致等。电池组不一致性是影响电动汽车使用性能和安全性的重要因素，为避免不一致引起的热失控和延缓不一致扩展，需在动力电池组使用过程中，对其一致性进行评估。Lithium-ion batteries have gradually become the most widely used power battery type for electric vehicles due to their high energy density, long cycle life, and low self-discharge rate. In order to meet the long-range and high-power requirements of electric vehicles, the on-board lithium-ion power battery pack is usually composed of several battery cells in series and parallel. Due to differences in manufacturing and use processes, a large number of battery cells in a vehicle-mounted power battery pack inevitably have inconsistent performance problems, which have various characteristics, such as inconsistent capacity, inconsistent internal resistance, and inconsistency in voltage. The inconsistency of the battery pack is an important factor affecting the performance and safety of electric vehicles. In order to avoid thermal runaway caused by inconsistencies and delay the expansion of inconsistencies, the consistency of the power battery pack needs to be evaluated during the use of the power battery pack.

现有电动汽车动力电池组一致性的评估方法主要有单参数评估法、多参数评估法等。单参数评估法主要是通过实验手段测定电池单体的容量、内阻以及SOC等参数，并计算其标准差、极差等统计量，从而表征电池组不一致性。这些参数的计算一般需要用到精密的测量仪器，并需要在特定的环境和工况下进行。近年来，有学者在文献中提出了多参数评估法，例如，利用熵权重法对动力电池组的容量、内阻等参数赋予不同的权重，进而评估电池组不一致性，或通过二维图形对电池的二参数不一致性进行评估等。The existing evaluation methods for the consistency of electric vehicle power battery packs mainly include single-parameter evaluation method and multi-parameter evaluation method. The single-parameter evaluation method mainly uses experimental methods to determine the capacity, internal resistance, and SOC of a battery cell, and calculates its standard deviation, range and other statistics to characterize the inconsistency of the battery pack. The calculation of these parameters generally requires the use of sophisticated measuring instruments, and needs to be carried out under specific environments and working conditions. In recent years, some scholars have proposed a multi-parameter evaluation method in the literature. For example, the entropy weight method is used to assign different weights to the parameters such as the capacity and internal resistance of the power battery pack to evaluate the inconsistency of the battery pack, or use a two-dimensional graph to The inconsistency of the two parameters of the battery is evaluated and so on.

现有电池组一致性评估技术存在的不足主要有以下几点：The main deficiencies of the existing battery pack consistency assessment technology are as follows:

(1)评估所需参数无法在线测量。现有文献和专利中的一致性评估方法所采用的容量、SOC、恒流恒压充电时间比等参数，需要精密的实验器材，而且需要特定的充放电工况，在电动汽车应用过程中难以测量，导致无法进行实际的应用。(1) The parameters required for evaluation cannot be measured online. The capacity, SOC, constant current and constant voltage charging time ratio and other parameters used in the consistency evaluation methods in the existing literature and patents require sophisticated experimental equipment and specific charging and discharging conditions, which are difficult to apply in the application process of electric vehicles. Measurements make it impossible to carry out practical applications.

(2)评估参数单一。电动汽车动力电池组在充电、放电以及静置工况下的特性不同，导致电池组的不一致性具有多参数耦合特性。因此，仅通过单一参数无法综合表征电池组多参数耦合不一致特性，而通过简单的加权方式对多参数不一致进行综合，存在评估不准确的问题。(2) The evaluation parameter is single. Electric vehicle power battery packs have different characteristics under charging, discharging and standing conditions, which leads to inconsistencies in battery packs with multi-parameter coupling characteristics. Therefore, only a single parameter cannot comprehensively characterize the inconsistent characteristics of the multi-parameter coupling of the battery pack, and the inconsistency of the multi-parameters is synthesized through a simple weighting method, and there is a problem of inaccurate evaluation.

发明内容Summary of the invention

基于此，本发明的目的是提供一种电池组一致性评估方法及***，以提高评估结果的准确性。Based on this, the purpose of the present invention is to provide a battery pack consistency evaluation method and system to improve the accuracy of the evaluation result.

为实现上述目的，本发明提供了一种电池组一致性评估方法，所述方法包括：To achieve the above objective, the present invention provides a battery pack consistency evaluation method, the method includes:

步骤S1：对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段；Step S1: Divide the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment;

步骤S2：基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压；Step S2: Extract the open circuit voltage of the battery cell from the battery cell voltage based on the fully charged static segment;

步骤S3：基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻；Step S3: Based on the driving segment, establish an equivalent circuit model for the battery cells, use the adaptive OCV-RLS method to identify the parameters of the battery cells, and extract the ohmic internal resistance of each battery cell;

步骤S4：基于所述恒流充电片段，计算各电池单体的充电电压向量范数；Step S4: Calculate the charging voltage vector norm of each battery cell based on the constant current charging segment;

步骤S5：根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性；Step S5: According to the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, combining Mahalanobis distance and DBSCAN clustering algorithm to calculate the consistency of each battery pack under test at each evaluation point;

步骤S6：根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数；Step S6: Calculate the Z score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point;

步骤S7：根据Z分数确定待测电池组的一致性。Step S7: Determine the consistency of the battery pack to be tested according to the Z score.

可选的，所述基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻，具体包括：Optionally, based on the driving segment, an equivalent circuit model is established for the battery cells, and the adaptive OCV-RLS method is used to identify the parameters of the battery cells, and the ohmic internal resistance of each battery cell is extracted. include:

步骤S31：确定待辨识参数估计公式；Step S31: Determine the parameter estimation formula to be identified;

步骤S32：采用差分电压方法确定初始时间步的电池开路电压估计值；Step S32: Use the differential voltage method to determine the estimated value of the battery open circuit voltage at the initial time step;

步骤S33：给定初始时间步的误差协方差矩阵和待辨识参数矩阵；Step S33: Given the error covariance matrix of the initial time step and the parameter matrix to be identified;

步骤S34：将第k-1时间步的误差协方差矩阵和待辨识参数矩阵代入所述待辨识参数估计公式，确定第k时间步的待辨识参数矩阵估计和误差协方差矩阵；Step S34: Substitute the error covariance matrix and the to-be-identified parameter matrix of the k-1 time step into the to-be-identified parameter estimation formula, and determine the to-be-identified parameter matrix estimation and the error covariance matrix of the k-th time step;

步骤S35：根据第k时间步的待辨识参数矩阵计算第k时间步的电池开路电压估计值；Step S35: Calculate the estimated value of the open circuit voltage of the battery at the k-th time step according to the to-be-identified parameter matrix at the k-th time step;

步骤S36：判断第k时间步的电池开路电压估计值是否满足电压异常判断规则；如果满足电压异常判断规则，则执行步骤S37；如果不满足电压异常判断规则，则执行步骤S38；所述电压异常判断规则为：定义电池放电电流为正，充电电流为负，当电池放电电流I _k为正时，电池开路电压估计值小于等于此刻的端电压，或大于前一时刻的电池开路电压估计值；若电池放电电流I _k为负时，电池开路电压估计值大于等于此刻的端电压，或小于前一时刻的电池开路电压估计值； Step S36: Determine whether the estimated value of the battery open circuit voltage at the kth time step meets the voltage abnormality judgment rule; if the voltage abnormality judgment rule is satisfied, go to step S37; if the voltage abnormality judgment rule is not met, go to step S38; the voltage is abnormal The judgment rule is: define the battery discharge current to be positive and the charge current to be negative. When the battery discharge current I _k is positive, the estimated value of the battery open circuit voltage is less than or equal to the terminal voltage at this moment, or greater than the estimated value of the battery open circuit voltage at the previous moment; If the battery discharge current I _k is negative, the estimated value of the battery open circuit voltage is greater than or equal to the terminal voltage at this moment, or less than the estimated value of the battery open circuit voltage at the previous moment;

步骤S37：采用差分电压方法确定第k时间步的电池开路电压估计值，根据第k时间步的电池开路电压估计值更新待辨识参数矩阵估计值，并执行步骤S38；Step S37: Determine the estimated value of the open circuit voltage of the battery at the k-th time step by using the differential voltage method, update the estimated value of the parameter matrix to be identified according to the estimated value of the open-circuit voltage of the battery at the k-th time step, and execute step S38;

步骤S38：判断k是否小于或等于时间步总数n；如果k小于或等于n，则令k＝k+1，并返回步骤S34；如果k大于n，则将辨识得到的每个电池单体的欧姆内阻输出。Step S38: Judge whether k is less than or equal to the total number of time steps n; if k is less than or equal to n, set k=k+1, and return to step S34; if k is greater than n, the identification of each battery cell Ohm resistance output.

可选的，所述采用差分电压方法确定第k时间步的电池开路电压估计值，具体包括：Optionally, the use of a differential voltage method to determine the estimated value of the battery open circuit voltage at the k-th time step specifically includes:

步骤S371：计算两个数据帧之间的电压差和电流差；Step S371: Calculate the voltage difference and the current difference between the two data frames;

步骤S372：以设定帧数据为一组，利用多个电压差和电流差构成的二维散点图，并对所述二维散点图进行线性回归分析，获得线性回归直线；Step S372: Taking the set frame data as a group, using a two-dimensional scatter diagram composed of multiple voltage differences and current differences, and performing linear regression analysis on the two-dimensional scatter diagram to obtain a linear regression line;

步骤S373：计算所述线性回归直线的斜率，并将所述斜率作为电池单体的内阻；Step S373: Calculate the slope of the linear regression line, and use the slope as the internal resistance of the battery cell;

步骤S374：基于所述电池单体的内阻计算电池开路电压估计值。Step S374: Calculate the estimated value of the open circuit voltage of the battery based on the internal resistance of the battery cell.

可选的，所述根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性，具体包括：Optionally, according to the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, the Mahalanobis distance and the DBSCAN clustering algorithm are combined to calculate the consistency of each battery pack under test at each evaluation point, which specifically includes :

步骤S51：基于设定时间内所有车辆各电池单体对应的开路电压、欧姆内阻和充电电压向量范数构成第一评估参数矩阵；Step S51: Construct a first evaluation parameter matrix based on the open circuit voltage, ohmic internal resistance and charging voltage vector norm corresponding to each battery cell of all vehicles within the set time;

步骤S52：采用DBSCAN聚类算法，根据所述第一评估参数矩阵删除离群的电池单体，并获得第二评估参数矩阵；Step S52: Use the DBSCAN clustering algorithm to delete outlier battery cells according to the first evaluation parameter matrix, and obtain a second evaluation parameter matrix;

步骤S53：根据所述第二评估参数矩阵计算剩余电池单体的协方差矩阵；Step S53: Calculate the covariance matrix of the remaining battery cells according to the second evaluation parameter matrix;

步骤S54：在每一个评估点，根据待测车辆的当前评估参数矩阵，采用DBSCAN聚类算法删除离群的电池单体，并获得第三评估参数矩阵；Step S54: At each evaluation point, according to the current evaluation parameter matrix of the vehicle to be tested, use the DBSCAN clustering algorithm to delete outlier battery cells, and obtain a third evaluation parameter matrix;

步骤S55：根据所述第三评估参数矩阵计算在m个参数上的平均值，获得数据中心点；Step S55: Calculate the average value of the m parameters according to the third evaluation parameter matrix to obtain the data center point;

步骤S56：基于所述协方差矩阵与所述数据中心点确定各电池单体与数据中心点之间的马氏距离；Step S56: Determine the Mahalanobis distance between each battery cell and the data center point based on the covariance matrix and the data center point;

步骤S57：基于各电池单体与数据中心点之间的马氏距离计算各评估点处各待测电池组的一致性。Step S57: Calculate the consistency of each battery pack to be tested at each evaluation point based on the Mahalanobis distance between each battery cell and the data center point.

可选的，所述对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段，具体包括：Optionally, the dividing the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment specifically includes:

步骤S11：从大数据平台采集的车辆时间序列数据，所述车辆时间序列数据包括车辆速度、经纬度、SOC、电池组总电压、总电流和单体电压；Step S11: vehicle time series data collected from a big data platform, the vehicle time series data including vehicle speed, latitude and longitude, SOC, total battery voltage, total current and cell voltage;

步骤S12：基于所述车辆时间序列数据选取满足满电静置要求的连续时间序列片段作为满电静置片段；所述满电静置要求为连续1小时以上速度为零、电流为零，且SOC等于100％的数据点；Step S12: Based on the vehicle time series data, a continuous time sequence segment that meets the requirements of full-charge standing is selected as a fully-charged standing segment; the full-charge standing requirement is that the speed is zero and the current is zero for more than 1 hour, and Data points where SOC is equal to 100%;

步骤S13：基于所述车辆时间序列数据选取满足行驶要求的连续时间序列片段作为行驶片段；所述行驶要求为第一特定SOC区间的行驶数据片段；Step S13: Based on the vehicle time series data, a continuous time sequence segment that meets the driving requirement is selected as a driving segment; the driving requirement is a driving data segment of a first specific SOC interval;

步骤S14：基于所述车辆时间序列数据选取满足恒流充电要求的连续时间序列片段作为恒流充电片段；所述恒流充电要求为在第二特定SOC区间内充电电流为恒定值。Step S14: Based on the vehicle time series data, a continuous time series segment that meets the constant current charging requirement is selected as the constant current charging segment; the constant current charging requirement is that the charging current is a constant value in the second specific SOC interval.

本发明还提供一种电池组一致性评估***，所述***包括：The present invention also provides a battery pack consistency evaluation system, which includes:

状态划分模块，用于对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段；The state division module is used to divide the state of the vehicle to obtain the fully charged stationary segment, the driving segment and the constant current charging segment;

开路电压确定模块，用于基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压；The open circuit voltage determination module is configured to extract the open circuit voltage of the battery cell from the battery cell voltage based on the fully charged static segment;

欧姆内阻确定模块，用于基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻；The ohmic resistance determination module is used to establish an equivalent circuit model for the battery cell based on the driving segment, use the adaptive OCV-RLS method to identify the battery cell parameters, and extract the ohmic resistance of each battery cell ；

范数确定模块，用于基于所述恒流充电片段，计算各电池单体的充电电压向量范数；A norm determining module, configured to calculate the norm of the charging voltage vector of each battery cell based on the constant current charging segment;

第一一致性确定模块，用于根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性；The first consistency determination module is used to calculate the consistency of each battery pack under test at each evaluation point based on the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, combined with Mahalanobis distance and DBSCAN clustering algorithm sex;

Z分数确定模块，用于根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数；The Z-score determination module is used to calculate the Z-score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point;

第二一致性确定模块，用于根据Z分数确定待测电池组的一致性。The second consistency determination module is used to determine the consistency of the battery pack to be tested according to the Z score.

可选的，所述欧姆内阻确定模块，具体包括：Optionally, the ohmic internal resistance determination module specifically includes:

待辨识参数估计公式确定单元，用于确定待辨识参数估计公式；The parameter estimation formula determination unit to be identified is used to determine the parameter estimation formula to be identified;

初始时间步的电池开路电压估计值确定单元，用于采用差分电压方法确定初始时间步的电池开路电压估计值；The battery open circuit voltage estimation value determination unit at the initial time step is used to determine the battery open circuit voltage estimation value at the initial time step by using a differential voltage method;

给定单元，用于给定初始时间步的误差协方差矩阵和待辨识参数矩阵；A given unit is used for a given initial time step error covariance matrix and parameter matrix to be identified;

参数确定单元，用于将第k-1时间步的误差协方差矩阵和待辨识参数矩阵代入所述待辨识参数估计公式，确定第k时间步的待辨识参数矩阵估计和误差协方差矩阵；The parameter determination unit is configured to substitute the error covariance matrix and the parameter matrix to be identified at the k-1 time step into the parameter estimation formula to be identified, and determine the parameter matrix estimation and error covariance matrix to be identified at the k time step;

第k时间步的电池开路电压估计值确定单元，用于根据第k时间步的待辨识参数矩阵计算第k时间步的电池开路电压估计值；The battery open-circuit voltage estimated value determination unit at the k-th time step is configured to calculate the battery open-circuit voltage estimated value at the k-th time step according to the to-be-identified parameter matrix at the k-th time step;

判断单元，用于判断第k时间步的电池开路电压估计值是否满足电压异常判断规则；如果满足电压异常判断规则，则执行“更新单元”；如果不满足电压异常判断规则，则执行“第二判断单元”；所述电压异常判断规则为：定义电池放电电流为正，充电电流为负，当电池放电电流I _k为正时，电池开路电压估计值小于等于此刻的端电压，或大于前一时刻的电池开路电压估计值；若电池放电电流I _k为负时，电池开路电压估计值大于等于此刻的端电压，或小于前一时刻的电池开路电压估计值； The judgment unit is used to judge whether the estimated value of the battery open circuit voltage at the kth time step meets the voltage abnormality judgment rule; if it meets the voltage abnormality judgment rule, execute the "update unit"; if it does not meet the voltage abnormality judgment rule, execute the "second Judgment unit"; The voltage abnormality judgment rule is: define the battery discharge current as positive and the charging current as negative. When the battery discharge current I _k is positive, the estimated value of the battery open circuit voltage is less than or equal to the terminal voltage at the moment, or greater than the previous one The estimated value of the battery open circuit voltage at the moment; if the battery discharge current I _k is negative, the estimated value of the battery open circuit voltage is greater than or equal to the terminal voltage at the moment, or less than the estimated value of the battery open circuit voltage at the previous moment;

更新单元，用于采用差分电压方法确定第k时间步的电池开路电压估计值，根据第k时间步的电池开路电压估计值更新待辨识参数矩阵估计值，并执行“第二判断单元”；The update unit is used to determine the estimated value of the open circuit voltage of the battery at the k-th time step by using the differential voltage method, update the estimated value of the parameter matrix to be identified according to the estimated value of the open-circuit voltage of the battery at the k-th time step, and execute the "second judgment unit";

第二判断单元，用于判断k是否小于或等于时间步总数n；如果k小于或等于n，则令k＝k+1，并返回“参数确定单元”；如果k大于n，则将辨识得到的每个电池单体的欧姆内阻输出。The second judgment unit is used to judge whether k is less than or equal to the total number of time steps n; if k is less than or equal to n, set k=k+1 and return to the "parameter determination unit"; if k is greater than n, it will be identified The ohmic internal resistance output of each battery cell.

可选的，所述更新单元，具体包括：Optionally, the update unit specifically includes:

电压差和电流差确定子单元，用于计算两个数据帧之间的电压差和电流差；The voltage difference and current difference determining subunit is used to calculate the voltage difference and current difference between two data frames;

线性回归分析子单元，用于以设定帧数据为一组，利用多个电压差和电流差构成的二维散点图，并对所述二维散点图进行线性回归分析，获得线性回归直线；The linear regression analysis subunit is used to set the frame data as a group, use a two-dimensional scatter diagram formed by multiple voltage differences and current differences, and perform linear regression analysis on the two-dimensional scatter diagram to obtain a linear regression straight line;

斜率计算确定子单元，用于计算所述线性回归直线的斜率，并将所述斜率作为电池单体的内阻；A slope calculation and determination subunit, used to calculate the slope of the linear regression line, and use the slope as the internal resistance of the battery cell;

电池开路电压估计值确定子单元，用于基于所述电池单体的内阻计算电池开路电压估计值；A battery open circuit voltage estimation value determination subunit, which is used to calculate the battery open circuit voltage estimation value based on the internal resistance of the battery cell;

更新子单元，用于根据电池开路电压估计值更新待辨识参数矩阵估计值。The update subunit is used to update the estimated value of the parameter matrix to be identified according to the estimated value of the open circuit voltage of the battery.

可选的，所述第一一致性确定模块，具体包括：Optionally, the first consistency determining module specifically includes:

第一评估参数矩阵确定单元，用于基于设定时间内所有车辆各电池单体对应的开路电压、欧姆内阻和充电电压向量范数构成第一评估参数矩阵；The first evaluation parameter matrix determination unit is configured to form a first evaluation parameter matrix based on the open circuit voltage, ohmic internal resistance and charging voltage vector norm corresponding to each battery cell of all vehicles within a set time;

第二评估参数矩阵确定单元，用于采用DBSCAN聚类算法，根据所述第一评估参数矩阵删除离群的电池单体，并获得第二评估参数矩阵；The second evaluation parameter matrix determination unit is configured to use the DBSCAN clustering algorithm to delete outlier battery cells according to the first evaluation parameter matrix, and obtain a second evaluation parameter matrix;

协方差矩阵确定单元，用于根据所述第二评估参数矩阵计算剩余电池单体的协方差矩阵；A covariance matrix determining unit, configured to calculate the covariance matrix of the remaining battery cells according to the second evaluation parameter matrix;

第三评估参数矩阵确定单元，用于在每一个评估点，根据待测车辆的当前评估参数矩阵，采用DBSCAN聚类算法删除离群的电池单体，并获得第三评估参数矩阵；The third evaluation parameter matrix determination unit is used to use the DBSCAN clustering algorithm to delete outlier battery cells at each evaluation point according to the current evaluation parameter matrix of the vehicle to be tested, and obtain the third evaluation parameter matrix;

数据中心点确定单元，用于根据所述第三评估参数矩阵计算在m个参数上的平均值，获得数据中心点；A data center point determining unit, configured to calculate an average value on m parameters according to the third evaluation parameter matrix to obtain a data center point;

马氏距离确定单元，用于基于所述协方差矩阵与所述数据中心点确定各电池单体与数据中心点之间的马氏距离；A Mahalanobis distance determining unit, configured to determine the Mahalanobis distance between each battery cell and the data center point based on the covariance matrix and the data center point;

第一一致性确定单元，用于基于各电池单体与数据中心点之间的马氏距离计算各评估点处各待测电池组的一致性。The first consistency determining unit is used to calculate the consistency of each battery pack to be tested at each evaluation point based on the Mahalanobis distance between each battery cell and the data center point.

可选的，所述状态划分模块，具体包括：Optionally, the state division module specifically includes:

车辆时间序列数据获取单元，用于从大数据平台采集的车辆时间序列数据，所述车辆时间序列数据包括车辆速度、经纬度、SOC、电池组总电压、总电流和单体电压；A vehicle time series data acquisition unit for vehicle time series data collected from a big data platform, the vehicle time series data including vehicle speed, latitude and longitude, SOC, total battery pack voltage, total current, and cell voltage;

满电静置片段确定单元，用于基于所述车辆时间序列数据选取满足满电静置要求的连续时间序列片段作为满电静置片段；所述满电静置要求为连续1小时以上速度为零、电流为零，且SOC等于100％的数据点；A fully charged stationary segment determining unit, configured to select a continuous time sequence segment that meets the requirement of fully charged stationary as a fully charged stationary segment based on the vehicle time series data; the requirement of fully charged stationary is that the speed of the fully charged stationary is more than 1 hour continuously. Zero, the current is zero, and the SOC is equal to the data point of 100%;

行驶片段确定单元，用于基于所述车辆时间序列数据选取满足行驶要求的连续时间序列片段作为行驶片段；所述行驶要求为第一特定SOC区间的行驶数据片段；A driving segment determining unit, configured to select a continuous time series segment that meets the driving requirement as a driving segment based on the vehicle time series data; the driving requirement is a driving data segment of a first specific SOC interval;

恒流充电片段确定单元，用于基于所述车辆时间序列数据选取满足恒流充电要求的连续时间序列片段作为恒流充电片段；所述恒流充电要求为在第二特定SOC区间内充电电流为恒定值。The constant current charging segment determining unit is configured to select a continuous time series segment that meets the constant current charging requirement as the constant current charging segment based on the vehicle time series data; the constant current charging requirement is that the charging current in the second specific SOC interval is Constant value.

根据本发明提供的具体实施例，本发明公开了以下技术效果：According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

本发明提供了一种电池组一致性评估方法及***，方法包括：对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段；基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压；基于行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻；基于恒流充电片段，计算各电池单体的充电电压向量范数；根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性；根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数；根据Z分数确定待测电池组的一致性。本发明基于自适应OCV-RLS方法实现对异常OCV辨识值进行补偿，进而提高电池组一致性评估结果的准确性。The present invention provides a battery pack consistency evaluation method and system. The method includes: dividing the state of the vehicle to obtain a fully charged static segment, a driving segment, and a constant current charging segment; based on the fully charged static segment, pass the battery The cell voltage extracts the open circuit voltage of the battery cell; based on the driving segment, the equivalent circuit model is established for the battery cell, and the adaptive OCV-RLS method is used to identify the parameters of the battery cell, and extract the ohm of each battery cell Resistance; Based on the constant current charging segment, calculate the charge voltage vector norm of each battery cell; According to the open circuit voltage, ohmic resistance and charge voltage vector norm of each battery cell, combine Mahalanobis distance and DBSCAN clustering algorithm to calculate each Evaluate the consistency of each battery pack to be tested at each evaluation point; calculate the Z score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point; determine the consistency of the battery pack to be tested based on the Z score. The present invention is based on the adaptive OCV-RLS method to realize the compensation of the abnormal OCV identification value, thereby improving the accuracy of the consistency evaluation result of the battery pack.

说明书附图Attached drawings

为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the present invention. For the embodiments, for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative labor.

图1为本发明实施例电池组一致性评估方法流程图；FIG. 1 is a flowchart of a method for evaluating the consistency of a battery pack according to an embodiment of the present invention;

图2为本发明实施例各评估点处各待测电池组的一致性评估结果示意图；2 is a schematic diagram of the consistency evaluation result of each battery pack to be tested at each evaluation point of the embodiment of the present invention;

图3为本发明实施例Z分数一致性评估示意图；Fig. 3 is a schematic diagram of Z score consistency evaluation according to an embodiment of the present invention;

图4为本发明实施例等效电路模型；Fig. 4 is an equivalent circuit model of an embodiment of the present invention;

图5为本发明实施例RLS方法和自适应OCV-RLS方法对比图；FIG. 5 is a comparison diagram of the RLS method and the adaptive OCV-RLS method according to the embodiment of the present invention;

图6为本发明实施例自适应OCV-RLS流程图；FIG. 6 is a flowchart of adaptive OCV-RLS according to an embodiment of the present invention;

图7为本发明实施例马氏距离与欧氏距离的对比图；FIG. 7 is a comparison diagram of Mahalanobis distance and Euclidean distance according to an embodiment of the present invention;

图8为本发明实施例电池组一致性评估***结构图。Fig. 8 is a structural diagram of a battery pack consistency evaluation system according to an embodiment of the present invention.

具体实施方式detailed description

下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

本发明的目的是提供一种电池组一致性评估方法及***，以提高评估结果的准确性。The purpose of the present invention is to provide a battery pack consistency evaluation method and system to improve the accuracy of the evaluation result.

为使本发明的上述目的、特征和优点能够更加明显易懂，下面结合附图和具体实施方式对本发明作进一步详细的说明。In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the attached drawings and specific embodiments.

(1)BMS：电池管理***(Battery Management System，简称BMS)，一般包括电池状态估计、热管理、均衡等功能。(1) BMS: Battery Management System (Battery Management System, BMS for short), which generally includes functions such as battery state estimation, thermal management, and equalization.

(2)马氏距离：马哈拉诺比斯距离(Mahalanobis Distance)，是一种距离的度量，可以看作是欧氏距离的一种修正，修正了欧式距离中各个维度尺度不一致且相关的问题，常被用作评定数据之间相似度的指标。(2) Mahalanobis distance: Mahalanobis Distance is a measure of distance, which can be regarded as a correction of Euclidean distance, which corrects the inconsistent and related dimensions of Euclidean distance. Questions are often used as indicators to assess the similarity between data.

(3)DBSCAN：一种基于密度的非参数聚类方法，它将数据集划分为核心点、边界点和噪声点，进而进行聚类，它具有较好的离群点检测能力，适用于凸样本集和非凸样本集。(3) DBSCAN: A density-based non-parametric clustering method, which divides the data set into core points, boundary points and noise points, and then performs clustering. It has good outlier detection capabilities and is suitable for convex Sample set and non-convex sample set.

(4)SOC：电池单体荷电状态(State of Charge，简称SOC)，描述了电池的剩余电量，其值为电池在一定放电倍率下，剩余电量与相同条件下额定容量的比值。(4) SOC: State of Charge (SOC for short), which describes the remaining power of the battery, and its value is the ratio of the remaining power to the rated capacity under the same conditions at a certain discharge rate.

(5)OCV：开路电压(Open Circuit Voltage，简称OCV)，电化学平衡状态下，电池正负两极的电势差。(5) OCV: Open Circuit Voltage (OCV), the potential difference between the positive and negative poles of the battery in an electrochemical equilibrium state.

(6)NCV：充电电压向量范数(Norm of Charge Voltage，简称NCV)，充电工况下，一段时间内电池电压组成的充电电压向量范数。(6) NCV: Charge voltage vector norm (Norm of Charge Voltage, NCV for short), which is the norm of the charge voltage vector composed of battery voltage in a period of time under charging conditions.

(7)RLS：递推最小二乘算法(Recursive Least Squares algorithm，简称RLS)。(7) RLS: Recursive Least Squares algorithm (RLS for short).

如图1所示，本发明公开一种电池组一致性评估方法，所述方法包括：As shown in Figure 1, the present invention discloses a battery pack consistency evaluation method, the method includes:

步骤S1：对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段。Step S1: Divide the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment.

步骤S2：基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压。Step S2: Based on the fully charged static segment, extract the open circuit voltage of the battery cell from the battery cell voltage.

步骤S3：基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻；自适应OCV-RLS方法为自适应开路电压OCV-遗忘因子递推最小二乘法RLS相结合的方法。Step S3: Based on the driving segment, establish an equivalent circuit model for the battery cells, use the adaptive OCV-RLS method to identify the parameters of the battery cells, and extract the ohmic internal resistance of each battery cell; adaptive OCV- The RLS method is a combination of adaptive open circuit voltage OCV and forgetting factor recursive least squares RLS.

步骤S4：基于所述恒流充电片段，计算各电池单体的充电电压向量范数。Step S4: Calculate the charge voltage vector norm of each battery cell based on the constant current charging segment.

步骤S5：根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性。Step S5: Calculate the consistency of each battery pack under test at each evaluation point based on the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, combined with Mahalanobis distance and DBSCAN clustering algorithm.

步骤S6：根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数。Step S6: Calculate the Z score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point.

下面对各个步骤进行详细论述：Each step is discussed in detail below:

步骤S1：对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段，具体包括：Step S1: Divide the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment, which specifically include:

步骤S11：从大数据平台采集车辆时间序列数据，所述车辆时间序列数据包括车辆速度、经纬度、SOC、电池组总电压、总电流和单体电压。Step S11: Collect vehicle time series data from a big data platform. The vehicle time series data includes vehicle speed, latitude and longitude, SOC, total battery voltage, total current, and cell voltage.

步骤S12：基于所述车辆时间序列数据选取满足满电静置要求的连续时间序列片段作为满电静置片段；所述满电静置要求为连续1小时以上速度为零、电流为零，且SOC等于100％的数据点。Step S12: Based on the vehicle time series data, a continuous time sequence segment that meets the requirements of full-charge standing is selected as a fully-charged standing segment; the full-charge standing requirement is that the speed is zero and the current is zero for more than 1 hour, and The SOC is equal to 100% of the data point.

步骤S13：基于所述车辆时间序列数据选取满足行驶要求的连续时间序列片段作为行驶片段；所述行驶要求为第一特定SOC区间的行驶数据片段，第一特定SOC区间是按照车辆行驶状态数据的SOC分布确定。作为实施例，第一特定SOC区间可取为80％～50％。Step S13: Based on the vehicle time series data, a continuous time sequence segment that meets the driving requirement is selected as the driving segment; the driving requirement is a driving data segment of a first specific SOC interval, and the first specific SOC interval is based on the vehicle driving state data SOC distribution is determined. As an example, the first specific SOC interval may be 80%-50%.

步骤S14：基于所述车辆时间序列数据选取满足恒流充电要求的连续时间序列片段作为恒流充电片段；所述恒流充电要求为在第二特定SOC区间内充电电流为恒定值，第二特定SOC区间和恒定电流值的选取均由车辆充电状态电池SOC分布和电流分布确定的。作为实施例，第二特定SOC区间可取为60％～80％，恒定电流值可取为16A。Step S14: Based on the vehicle time series data, a continuous time series segment that meets the constant current charging requirement is selected as the constant current charging segment; the constant current charging requirement is that the charging current is a constant value in the second specific SOC interval, and the second specific The selection of the SOC interval and the constant current value are determined by the SOC distribution and current distribution of the battery in the state of charge of the vehicle. As an example, the second specific SOC interval may be 60% to 80%, and the constant current value may be 16A.

本发明提取满电静置片段、行驶片段以及恒流充电片段作为后续一致性评估的数据基础。The present invention extracts the fully charged static segment, the driving segment and the constant current charging segment as the data basis for subsequent consistency evaluation.

步骤S3：基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻IR。自适应开路电压的遗忘因子递推最小二乘法简称自适应OCV-RLS。Step S3: Based on the driving segment, establish an equivalent circuit model for the battery cells, use the adaptive OCV-RLS method to identify the parameters of the battery cells, and extract the ohmic internal resistance IR of each battery cell. The forgetting factor recursive least square method of adaptive open circuit voltage is abbreviated as adaptive OCV-RLS.

为在线辨识车辆电池组内阻不一致性情况，需要建立电池模型。考虑模型精度、计算复杂度以及实车应用可行性，综合对比分析经验模型、电化学模型以及等效电路模型，本发明最终采用戴维南等效电路单体电池模型，如图4所示。In order to identify the inconsistency of the internal resistance of the vehicle battery pack online, a battery model needs to be established. Considering model accuracy, computational complexity and practical vehicle application feasibility, comprehensively comparing and analyzing the empirical model, electrochemical model and equivalent circuit model, the present invention finally adopts the Thevenin equivalent circuit single cell model, as shown in FIG. 4.

然而，受模型协方差的累积效应和数据传输误差的限制，RLS方法时常难以正确估计等效电路模型中的开路电压OCV，表现为两种类型的故障，如图5所示。图5中的(a)部分，传统的RLS方法估计的OCV波动很大，在图5中的(b)部分，估计的OCV有一些缺失值，这会降低参数识别的稳定性。因此，引入了差分电压(DV)方法来补偿估计误差。However, limited by the cumulative effect of model covariance and data transmission errors, the RLS method is often difficult to correctly estimate the open circuit voltage OCV in the equivalent circuit model, which manifests as two types of faults, as shown in Figure 5. In part (a) of Fig. 5, the OCV estimated by the traditional RLS method fluctuates greatly. In part (b) of Fig. 5, the estimated OCV has some missing values, which will reduce the stability of parameter identification. Therefore, the differential voltage (DV) method is introduced to compensate the estimation error.

图6为本发明实施例自适应OCV-RLS流程图，如图6所示，步骤S3具体包括：Fig. 6 is a flowchart of adaptive OCV-RLS according to an embodiment of the present invention. As shown in Fig. 6, step S3 specifically includes:

步骤S31：确定待辨识参数估计公式，具体包括：Step S31: Determine the parameter estimation formula to be identified, which specifically includes:

步骤S311：根据基尔霍夫定律，构建等效电路模型的动态方程，具体公式为：Step S311: According to Kirchhoff's law, construct the dynamic equation of the equivalent circuit model. The specific formula is:

其中，U _t(t)表示第t时刻电池端电压，I(t)表示第t时刻电池电流，R ₀表示欧姆内阻，U _OCV(t)表示第t时刻电池开路电压，R _P表示电池极化内阻，C _p表示极化电容，U _P(t)表示第t时刻极化电压。 Among them, U _t (t) represents battery terminal voltage at time t, I(t) represents battery current at time t, R ₀ represents ohmic resistance, U _OCV (t) represents battery open circuit voltage at time t, and R _P represents battery Polarization internal resistance, C _p represents the polarization capacitance, U _P (t) represents the polarization voltage at time t.

步骤S312：将动态方程转化为状态空间方程，并进行离散化，获得离散方程，具体公式为：Step S312: Convert the dynamic equation into a state space equation and perform discretization to obtain a discrete equation. The specific formula is:

其中，T表示采样间隔，U _t,k表示第k时间步的电池端电压，I _k表示第k时间步的电池电流，U _ocv,k表示第k时间步的电池开路电压，下角标k表示离散时间序号，也称为时间步。 Among them, T represents the sampling interval, U _t,k represents the battery terminal voltage at _{the k-th time step, I k} represents the battery current _{at the k-th} time step, U ocv,k represents the battery open-circuit voltage at the k-th time step, and the subscript k represents Discrete time sequence number, also called time step.

步骤S313：简化离散方程，具体公式为：Step S313: Simplify the discrete equation, the specific formula is:

其中，

表示输入矩阵，

θ _k表示待辨识参数矩阵，θ _k＝[α ₁ α ₂ α ₃ α ₄] ^T，

in,

Represents the input matrix,

θ _k represents the parameter matrix to be identified, θ _k =[α ₁ α ₂ α ₃ α ₄ ] ^T ,

步骤S314：基于简化后的离散方程，采用具有遗忘因子的递推最小二乘法(RLS)确定待辨识参数估计公式，具体公式为：Step S314: Based on the simplified discrete equation, the recursive least squares method (RLS) with forgetting factor is used to determine the parameter estimation formula to be identified. The specific formula is:

其中，μ表示遗忘因子，K _k表示算法的增益，P _k表示误差协方差矩阵，y _k表示端电压，

表示待辨识参数矩阵估计值，

表示输入矩阵，下角标k为时间步。 Among them, μ represents the forgetting factor, K _k represents the gain of the algorithm, P _k represents the error covariance matrix, y _k represents the terminal voltage,

Indicates the estimated value of the parameter matrix to be identified,

Represents the input matrix, and the subscript k is the time step.

步骤S32：采用差分电压方法确定初始时间步的电池开路电压估计值。Step S32: Use the differential voltage method to determine the estimated value of the battery open circuit voltage at the initial time step.

步骤S33：给定初始时间步的误差协方差矩阵和待辨识参数矩阵。Step S33: Given the error covariance matrix of the initial time step and the parameter matrix to be identified.

步骤S34：将第k-1时间步的误差协方差矩阵和待辨识参数矩阵代入所述待辨识参数估计公式，确定第k时间步的待辨识参数矩阵估计和误差协方差矩阵；也就是说，当k为1时，则第k-1时间步为初始时间步。Step S34: Substitute the error covariance matrix and the to-be-identified parameter matrix of the k-1 time step into the to-be-identified parameter estimation formula, and determine the to-be-identified parameter matrix estimation and the error covariance matrix of the k-th time step; that is, When k is 1, the k-1 time step is the initial time step.

步骤S35：根据第k时间步的待辨识参数矩阵计算第k时间步的电池开路电压估计值。Step S35: Calculate the estimated value of the open circuit voltage of the battery at the k-th time step according to the to-be-identified parameter matrix at the k-th time step.

步骤S36：判断第k时间步的电池开路电压估计值是否满足电压异常判断规则；如果满足电压异常判断规则，则执行步骤S37；如果不满足电压异常判断规则，则执行步骤S38。电压异常判断规则为：定义电池放电电流为正，充电电流为负，当电池放电电流I _k为正时，电池开路电压估计值小于等于此刻的端电压，或大于前一时刻的电池开路电压估计值；若电池放电电流Ik为负时，电池开路电压估计值大于等于此刻的端电压，或小于前一时刻的电池开路电压估计值。本发明当电流为正时，其端电压应该低于开路电压；而当电流为负时，端电压应该高于开路电压。 Step S36: Determine whether the estimated value of the battery open circuit voltage at the k-th time step meets the voltage abnormality judgment rule; if the voltage abnormality judgment rule is satisfied, step S37 is executed; if the voltage abnormality judgment rule is not satisfied, then step S38 is executed. The rule for judging abnormal voltage is: define battery discharge current as positive and charge current as negative. When battery discharge current I _k is positive, the estimated value of battery open circuit voltage is less than or equal to the terminal voltage at this moment, or greater than the estimated battery open circuit voltage at the previous moment. Value; if the battery discharge current Ik is negative, the estimated value of the battery open circuit voltage is greater than or equal to the terminal voltage at the moment, or less than the estimated value of the battery open circuit voltage at the previous moment. In the present invention, when the current is positive, the terminal voltage should be lower than the open circuit voltage; when the current is negative, the terminal voltage should be higher than the open circuit voltage.

步骤S37：采用差分电压方法确定第k时间步的电池开路电压估计值，根据第k时间步的电池开路电压估计值更新待辨识参数矩阵估计值，并执行步骤S38。Step S37: Determine the estimated value of the open circuit voltage of the battery at the k-th time step by using the differential voltage method, update the estimated value of the parameter matrix to be identified according to the estimated value of the open-circuit voltage of the battery at the k-th time step, and execute step S38.

步骤S37：采用差分电压方法确定第k时间步的电池开路电压估计值，具体包括：Step S37: Use the differential voltage method to determine the estimated value of the open-circuit voltage of the battery at the k-th time step, which specifically includes:

步骤S371：计算两个数据帧之间的电压差ΔU和电流差ΔI。Step S371: Calculate the voltage difference ΔU and the current difference ΔI between the two data frames.

步骤S372：以设定帧数据为一组，利用多个电压差ΔU和电流差ΔI构成的二维散点图，并对二维散点图进行线性回归分析，获得线性回归直线。Step S372: Taking the set frame data as a group, using a two-dimensional scatter plot composed of multiple voltage differences ΔU and current differences ΔI, and performing linear regression analysis on the two-dimensional scatter plot to obtain a linear regression line.

步骤S373：计算线性回归直线的斜率，并将斜率作为电池单体的内阻。Step S373: Calculate the slope of the linear regression line, and use the slope as the internal resistance of the battery cell.

步骤S374：基于电池单体的内阻计算电池开路电压估计值，具体公式为：Step S374: Calculate the estimated value of the open circuit voltage of the battery based on the internal resistance of the battery cell. The specific formula is:

其中，

表示利用DV法估算的第k时间步的电池开路电压估计值，R表示电池单体的内阻，I _k表示第k时间步的电池电流，U _t,k表示第k时间步的电池端电压。 in,

Represents the estimated value of the open circuit voltage of the battery at the k-th time step estimated by the DV method, R represents the internal resistance of the battery cell, I _k represents the battery current _{at the k-th} time step, U t,k represents the battery terminal voltage at the k-th time step .

步骤S4：基于所述恒流充电片段，计算各电池单体的充电电压向量范数NCV，具体公式为：Step S4: Calculate the charge voltage vector norm NCV of each battery cell based on the constant current charging segment, and the specific formula is:

NCV _i＝norm(v _i)＝||(v _i,1,v _i,2,...,v _i,t)||； NCV _i =norm(v _i )=||(v _i,1 ,vi _,2 ,...,vi _,t )||;

其中，v _i,t表示第i个电池单体在第t时刻的电压，NCV _i表示第i个电池单体的充电电压向量范数。 Among them, v _i,t represents the voltage of the i-th battery cell at time t, and NCV _i represents the charging voltage vector norm of the i-th battery cell.

本发明计算各电池单体在第1监测点到第t监测点的充电电压向量范数(NCV)，以表征单体充电性能。作为实施例，t可选取200，也可根据实际数据情况增大或减小。The invention calculates the charge voltage vector norm (NCV) of each battery cell from the first monitoring point to the t-th monitoring point to characterize the charging performance of the cell. As an example, t can be selected as 200, and can also be increased or decreased according to actual data conditions.

对于一个多维离散点集，评估其离散程度的传统方法为求其标准差，即先计算其中心点位置，再计算每个点到中心点的欧式距离，并求其标准差，用来表征离散程度，即一致性情况。然而，基于欧氏距离的方法的不足之处在于，它没有考虑不同参数的数量级和参数之间的相关性，即使对数据进行归一化，也无法解决某些方差较大的参数的变化可能会覆盖方差较小的参数的变化的问题。假设在二维空间R ²中的一组数据点具有椭圆型分布，如图7所示，C是数据集的中心点。在欧氏距离计算中，点E和点A到点C的距离相同，然而，我们可以从分布直观地看出，点E处于分布内部，而点A处于分布的边缘，更趋近于离群点。马氏距离即解决了上述问题。马氏距离考虑了数据不同属性的数量级以及其之间的相关性，将车辆时间序列数据集映射到主成分空间，再进行归一化处理。在马氏距离构成的空间中，点M到中心点的距离与A相同。 For a multi-dimensional discrete point set, the traditional method of evaluating the degree of dispersion is to find its standard deviation, that is, first calculate the position of its center point, and then calculate the Euclidean distance from each point to the center point, and find its standard deviation, which is used to characterize the dispersion Degree, that is, consistency. However, the disadvantage of the method based on Euclidean distance is that it does not consider the magnitude of different parameters and the correlation between the parameters. Even if the data is normalized, it cannot solve the possibility of changes in some parameters with large variances. Will cover the problem of changes in parameters with smaller variance. Assuming that a set of data points in the two-dimensional space R ² has an elliptical distribution, as shown in Figure 7, C is the center point of the data set. In the Euclidean distance calculation, the distance between point E and point A to point C is the same. However, we can intuitively see from the distribution that point E is inside the distribution and point A is at the edge of the distribution, which is closer to outlier point. The Mahalanobis distance solves the above problems. The Mahalanobis distance considers the magnitude of the different attributes of the data and the correlation between them, and maps the vehicle time series data set to the principal component space, and then normalizes it. In the space formed by the Mahalanobis distance, the distance from the point M to the center point is the same as that of A.

然而，由于整体数据协方差的差异，可能导致同样的一组样本数据在不同整体数据中的马氏距离计算出现不同，因此，本发明采用整体数据集的协方差矩阵来代替传统马氏距离计算中的协方差矩阵，以获得较为稳定的一致性评估。此外，马氏距离的计算中，协方差矩阵的计算以及中心点的选取，都对离群点较为敏感，异常点的存在会极大的影响评估结果。因此，本发明采用DBSCAN聚类算法对数据集中的离群点进行识别和剔除。DBSCAN聚类算法是一种基于密度的非参数聚类方法，它将数据集划分为核心点、边界点和噪声点，进而进行聚类，它具有较好的离群点检测能力。However, due to the difference in the overall data covariance, the Mahalanobis distance calculation of the same set of sample data in different overall data may be different. Therefore, the present invention uses the covariance matrix of the overall data set to replace the traditional Mahalanobis distance calculation. The covariance matrix in in order to obtain a more stable consistency evaluation. In addition, in the calculation of Mahalanobis distance, the calculation of the covariance matrix and the selection of the center point are all sensitive to outliers, and the existence of abnormal points will greatly affect the evaluation results. Therefore, the present invention uses the DBSCAN clustering algorithm to identify and eliminate outliers in the data set. The DBSCAN clustering algorithm is a density-based non-parametric clustering method. It divides the data set into core points, boundary points and noise points, and then performs clustering. It has good outlier detection capabilities.

步骤S5：根据各电池单体的开路电压OCV、欧姆内阻IR和充电电压向量范数NCV，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性，具体包括：Step S5: According to the open circuit voltage OCV, ohmic internal resistance IR and charging voltage vector norm NCV of each battery cell, combining Mahalanobis distance and DBSCAN clustering algorithm to calculate the consistency of each battery pack under test at each evaluation point, which specifically includes :

步骤S51：基于设定时间内所有车辆各电池单体对应的开路电压 OCV、欧姆内阻IR和充电电压向量范数NCV构成第一评估参数矩阵，具体公式为：Step S51: Based on the open circuit voltage OCV, the ohmic internal resistance IR and the charging voltage vector norm NCV corresponding to each battery cell of all vehicles within the set time, the first evaluation parameter matrix is formed. The specific formula is:

其中，X _total表示第一评估参数矩阵，n表示电池单体的总个数，m为一致性评估参数的个数，作为实施例，m可取值为3，即三个参数分别为开路电压(OCV)、欧姆内阻(IR)和充电电压的范数(NCV)，x _ij表示第i个电池单体的第j个评估参数值。 Among them, X _total represents the first evaluation parameter matrix, n represents the total number of battery cells, and m is the number of consistency evaluation parameters. As an example, m can take a value of 3, that is, the three parameters are open circuit voltages. (OCV), ohmic internal resistance (IR) and norm of charging voltage (NCV), x _ij represents the j-th evaluation parameter value of the i-th battery cell.

步骤S52：采用DBSCAN聚类算法，根据所述第一评估参数矩阵删除离群的电池单体，并获得第二评估参数矩阵X' _total。 Step S52: Using the DBSCAN clustering algorithm, delete outlier battery cells according to the first evaluation parameter matrix, and obtain a second evaluation parameter matrix _X'total .

步骤S53：根据所述第二评估参数矩阵计算剩余电池单体的协方差矩阵，具体公式为：Step S53: Calculate the covariance matrix of the remaining battery cells according to the second evaluation parameter matrix, the specific formula is:

其中，∑ _X,total表示剩余电池单体的协方差矩阵，X' _total表示第二评估参数矩阵，μ _X＝(μ ₁,μ ₂,...,μ _m)是一致性评估参数的均值矩阵，μ _m表示第m个一致性评估参数的均值，n表示电池单体的总个数。 Among them, ∑ _{X, total} represents the covariance matrix of the remaining battery cells, _X'total represents the second evaluation parameter matrix, μ _X = (μ ₁ , μ ₂ ,..., μ _m ) is the mean value of the consistency evaluation parameters Matrix, μ _m represents the mean value of the m-th consistency evaluation parameter, and n represents the total number of battery cells.

步骤S54：在每一个评估点，根据待测车辆的当前评估参数矩阵，采用DBSCAN聚类算法删除离群的电池单体，并获得第三评估参数矩阵，具体公式为：Step S54: At each evaluation point, according to the current evaluation parameter matrix of the vehicle to be tested, the DBSCAN clustering algorithm is used to delete outlier battery cells, and a third evaluation parameter matrix is obtained. The specific formula is:

其中，X表示第三评估参数矩阵，q表示待测车辆的电池单体的总个数，q≤n，m为一致性评估参数的个数。Among them, X represents the third evaluation parameter matrix, q represents the total number of battery cells of the vehicle to be tested, q≤n, and m is the number of consistency evaluation parameters.

步骤S55：根据所述第三评估参数矩阵计算在m个参数上的平均值，获得数据中心点。Step S55: Calculate the average value of the m parameters according to the third evaluation parameter matrix to obtain the data center point.

步骤S56：基于所述协方差矩阵与所述数据中心点确定各电池单体与数据中心点之间的马氏距离，具体公式为：Step S56: Determine the Mahalanobis distance between each battery cell and the data center point based on the covariance matrix and the data center point, the specific formula is:

其中，D _M(X _j)表示第j个电池单体与数据中心点之间的马氏距离，X _j为第j个电池单体，μ′ _X为数据中心点，∑ _X,total为协方差矩阵。 Among them, D _M (X _j ) represents the Mahalanobis distance between the j-th battery cell and the data center point, X _j is the j-th battery cell, μ′ _X is the data center point, and ∑ _X,total is the agreement Variance matrix.

步骤S57：基于各电池单体与数据中心点之间的马氏距离计算各评估点处各待测电池组的一致性，具体公式为：Step S57: Calculate the consistency of each battery pack to be tested at each evaluation point based on the Mahalanobis distance between each battery cell and the data center point. The specific formula is:

其中，D _M,total表示各评估点处各待测电池组的一致性，D _M(X _j)表示第j个电池单体与中心点之间的马氏距离。 Among them, D _M,total represents the consistency of each battery pack to be tested at each evaluation point, and D _M (X _j ) represents the Mahalanobis distance between the j-th battery cell and the center point.

例如以每隔1000公里为一个里程级别，从车辆全部数据中分别提取里程数与里程级别最相近的符合步骤S2、步骤S3、步骤S4的数据，聚合为一个一致性评估点。For example, taking every 1,000 kilometers as a mileage level, extract data from all vehicle data that are closest to the mileage level and comply with step S2, step S3, and step S4, and aggregate them into a consistency evaluation point.

图2为本发明的一个实施例中9辆电动汽车动力电池组一致性评估结果随里程的演变情况。可以看出，9辆电动汽车的一致性情况遵循相似的演变规律，里程5000km到11000km之间，一致性较好，5000km之前和11000km之后，一致性较差。这是因为5000km之前以及11000km之后处于冬季，温度较低，导致的车辆动力电池组一致性变差。而对比5000km之前以及11000km之后两个区间，发现动力电池组一致性出现了衰退。图中，5号车的电池一致性在113000km处发生了异常，9号车的电池一致性在141000km处发生了异常，它们的一致性评估结果相比其他车辆更差。Figure 2 shows the evolution of the consistency evaluation results of the power battery packs of 9 electric vehicles with mileage in an embodiment of the present invention. It can be seen that the consistency of the 9 electric vehicles follows a similar evolutionary law. The consistency is good between 5000km and 11000km, and the consistency is poor before 5000km and after 11000km. This is because in winter before 5000km and after 11000km, the temperature is low, resulting in poor consistency of the vehicle power battery pack. Comparing the two intervals before 5000km and after 11000km, it is found that the consistency of the power battery pack has declined. In the figure, the battery consistency of car 5 is abnormal at 113000km, and the battery consistency of car 9 is abnormal at 141000km, and their consistency evaluation results are worse than other vehicles.

步骤S6：根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数，具体公式为：Step S6: Calculate the Z score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point, the specific formula is:

其中，D _M,total表示评估点处待测电池组的一致性，μ表示评估点处所有待测电池组的一致性均值，σ表示评估点处的标准差。 Among them, D _M,total represents the consistency of the battery pack to be tested at the evaluation point, μ represents the average consistency of all battery packs to be tested at the evaluation point, and σ represents the standard deviation at the evaluation point.

步骤S7：根据Z分数确定待测电池组的一致性，具体包括：Step S7: Determine the consistency of the battery pack to be tested according to the Z score, which specifically includes:

当Z分数小于第一设定值时，表示车辆待测电池组内的各电池单体一致性较好。When the Z score is less than the first set value, it indicates that the battery cells in the battery pack to be tested have good consistency.

当第一设定值≤Z分数<第二设定值时，表示车辆待测电池组内的各电池单体出现了轻微的不一致。When the first set value≤Z score<the second set value, it means that the battery cells in the battery pack to be tested are slightly inconsistent.

当第二设定值≤Z分数<第三设定值时，表示车辆待测电池组内的各电池单体出现中等程度不一致，建议采取均衡等维护措施。When the second set value≤Z score<the third set value, it means that the battery cells in the battery pack to be tested are moderately inconsistent. It is recommended to take maintenance measures such as balancing.

当第三设定值≤Z分数时，表示车辆待测电池组内的各电池单体出现严重的不一致，报警信号启动。When the third set value ≤ Z score, it means that the battery cells in the vehicle battery pack to be tested are seriously inconsistent, and the alarm signal is activated.

图3为本发明的一个实施例中9辆电动汽车一致性Z分数评级，其更加清晰的显示出电池一致性较差的车辆。例如，在113000km处和141000km处，5号车和9号车的Z分数超过了3，达到第3级预警。Fig. 3 shows the consistency Z score ratings of 9 electric vehicles in an embodiment of the present invention, which more clearly shows vehicles with poor battery consistency. For example, at 113000km and 141000km, the Z scores of cars 5 and 9 exceed 3, reaching the level 3 warning.

如图8所示，本发明还提供一种电池组一致性评估***，所述***包括：As shown in Figure 8, the present invention also provides a battery pack consistency evaluation system, which includes:

状态划分模块1，用于对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段；The state division module 1 is used to divide the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment;

开路电压确定模块2，用于基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压.Open-circuit voltage determination module 2, for extracting the open-circuit voltage of the battery cell based on the fully charged static segment, by the battery cell voltage.

欧姆内阻确定模块3，用于基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻。The ohmic resistance determination module 3 is used to establish an equivalent circuit model for the battery cell based on the driving segment, use the adaptive OCV-RLS method to identify the battery cell parameters, and extract the ohmic resistance of each battery cell Hinder.

范数确定模块4，用于基于所述恒流充电片段，计算各电池单体的充电电压向量范数。The norm determining module 4 is configured to calculate the norm of the charging voltage vector of each battery cell based on the constant current charging segment.

第一一致性确定模块5，用于根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性。The first consistency determination module 5 is used to calculate the battery pack under test at each evaluation point based on the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, combined with Mahalanobis distance and DBSCAN clustering algorithm consistency.

Z分数确定模块6，用于根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数。The Z-score determining module 6 is used to calculate the Z-score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point.

第二一致性确定模块7，用于根据Z分数确定待测电池组的一致性。The second consistency determining module 7 is used to determine the consistency of the battery pack to be tested according to the Z score.

作为一种可选的实施方式，本发明所述欧姆内阻确定模块3，具体包括：As an optional implementation manner, the ohmic internal resistance determination module 3 of the present invention specifically includes:

待辨识参数估计公式确定单元，用于确定待辨识参数估计公式。The parameter estimation formula determination unit to be identified is used to determine the parameter estimation formula to be identified.

初始时间步的电池开路电压估计值确定单元，用于采用差分电压方法确定初始时间步的电池开路电压估计值。The battery open circuit voltage estimated value determining unit at the initial time step is used to determine the battery open circuit voltage estimated value at the initial time step by using a differential voltage method.

给定单元，用于给定初始时间步的误差协方差矩阵和待辨识参数矩阵。A given unit is used for a given initial time step error covariance matrix and parameter matrix to be identified.

参数确定单元，用于将第k-1时间步的误差协方差矩阵和待辨识参数矩阵代入所述待辨识参数估计公式，确定第k时间步的待辨识参数矩阵估计和误差协方差矩阵。The parameter determination unit is configured to substitute the error covariance matrix and the parameter matrix to be identified at the k-1 time step into the parameter estimation formula to be identified to determine the parameter matrix estimation and error covariance matrix to be identified at the k time step.

第k时间步的电池开路电压估计值确定单元，用于根据第k时间步的待辨识参数矩阵计算第k时间步的电池开路电压估计值。The battery open-circuit voltage estimated value determination unit at the k-th time step is configured to calculate the battery open-circuit voltage estimated value at the k-th time step according to the to-be-identified parameter matrix at the k-th time step.

判断单元，用于判断第k时间步的电池开路电压估计值是否满足电压异常判断规则；如果满足电压异常判断规则，则执行“更新单元”；如果不满足电压异常判断规则，则执行“第二判断单元”；所述电压异常判断规则为：定义电池放电电流为正，充电电流为负，当电池放电电流I _k为正时，电池开路电压估计值小于等于此刻的端电压，或大于前一时刻的电池开路电压估计值；若电池放电电流I _k为负时，电池开路电压估计值大于等于此刻的端电压，或小于前一时刻的电池开路电压估计值。 The judgment unit is used to judge whether the estimated value of the battery open circuit voltage at the kth time step meets the voltage abnormality judgment rule; if it meets the voltage abnormality judgment rule, execute the "update unit"; if it does not meet the voltage abnormality judgment rule, execute the "second Judgment unit"; The voltage abnormality judgment rule is: define the battery discharge current as positive and the charging current as negative. When the battery discharge current I _k is positive, the estimated value of the battery open circuit voltage is less than or equal to the terminal voltage at the moment, or greater than the previous one The estimated value of the battery open circuit voltage at the moment; if the battery discharge current I _k is negative, the estimated value of the battery open circuit voltage is greater than or equal to the terminal voltage at this moment, or less than the estimated value of the battery open circuit voltage at the previous moment.

更新单元，用于采用差分电压方法确定第k时间步的电池开路电压估计值，根据第k时间步的电池开路电压估计值更新待辨识参数矩阵估计值，并执行“第二判断单元”。The update unit is used to determine the estimated value of the open circuit voltage of the battery at the k-th time step by using the differential voltage method, update the estimated value of the parameter matrix to be identified according to the estimated value of the open-circuit voltage of the battery at the k-th time step, and execute the "second judgment unit".

作为一种可选的实施方式，本发明所述更新单元，具体包括：As an optional implementation manner, the update unit of the present invention specifically includes:

电压差和电流差确定子单元，用于计算两个数据帧之间的电压差和电流差。The voltage difference and current difference determining sub-unit is used to calculate the voltage difference and current difference between two data frames.

斜率计算确定子单元，用于计算所述线性回归直线的斜率，并将所述斜率作为电池单体的内阻。The slope calculation and determination subunit is used to calculate the slope of the linear regression line, and use the slope as the internal resistance of the battery cell.

作为一种可选的实施方式，本发明所述第一一致性确定模块5，具体包括：As an optional implementation manner, the first consistency determining module 5 of the present invention specifically includes:

第一评估参数矩阵确定单元，用于基于设定时间内所有车辆各电池单体对应的开路电压、欧姆内阻和充电电压向量范数构成第一评估参数矩阵。The first evaluation parameter matrix determination unit is configured to form a first evaluation parameter matrix based on the open circuit voltage, ohmic internal resistance and charging voltage vector norm corresponding to each battery cell of all vehicles within the set time.

第二评估参数矩阵确定单元，用于采用DBSCAN聚类算法，根据所述第一评估参数矩阵删除离群的电池单体，并获得第二评估参数矩阵。The second evaluation parameter matrix determination unit is configured to use the DBSCAN clustering algorithm to delete outlier battery cells according to the first evaluation parameter matrix, and obtain a second evaluation parameter matrix.

协方差矩阵确定单元，用于根据所述第二评估参数矩阵计算剩余电池单体的协方差矩阵。The covariance matrix determining unit is configured to calculate the covariance matrix of the remaining battery cells according to the second evaluation parameter matrix.

第三评估参数矩阵确定单元，用于在每一个评估点，根据待测车辆的当前评估参数矩阵，采用DBSCAN聚类算法删除离群的电池单体，并获得第三评估参数矩阵。The third evaluation parameter matrix determination unit is used for each evaluation point, according to the current evaluation parameter matrix of the vehicle to be tested, using the DBSCAN clustering algorithm to delete outlier battery cells, and obtain the third evaluation parameter matrix.

数据中心点确定单元，用于根据所述第三评估参数矩阵计算在m个参数上的平均值，获得数据中心点。The data center point determination unit is configured to calculate the average value of the m parameters according to the third evaluation parameter matrix to obtain the data center point.

马氏距离确定单元，用于基于所述协方差矩阵与所述数据中心点确定各电池单体与数据中心点之间的马氏距离。The Mahalanobis distance determining unit is configured to determine the Mahalanobis distance between each battery cell and the data center point based on the covariance matrix and the data center point.

作为一种可选的实施方式，本发明所述状态划分模块1，具体包括：As an optional implementation manner, the state division module 1 of the present invention specifically includes:

车辆时间序列数据获取单元，用于从大数据平台采集的车辆时间序列数据，所述车辆时间序列数据包括车辆速度、经纬度、SOC、电池组总电压、总电流和单体电压。The vehicle time series data acquisition unit is used to collect vehicle time series data from a big data platform, the vehicle time series data including vehicle speed, latitude and longitude, SOC, battery pack total voltage, total current and cell voltage.

满电静置片段确定单元，用于基于所述车辆时间序列数据选取满足满电静置要求的连续时间序列片段作为满电静置片段；所述满电静置要求为连续1小时以上速度为零、电流为零，且SOC等于100％的数据点。A fully charged stationary segment determining unit, configured to select a continuous time sequence segment that meets the requirement of fully charged stationary as a fully charged stationary segment based on the vehicle time series data; the requirement of fully charged stationary is that the speed of the fully charged stationary is more than 1 hour continuously. Zero, the current is zero, and the SOC is equal to the data point of 100%.

行驶片段确定单元，用于基于所述车辆时间序列数据选取满足行驶要求的连续时间序列片段作为行驶片段；所述行驶要求为第一特定SOC区间的行驶数据片段。The driving segment determination unit is configured to select a continuous time sequence segment that meets the driving requirement as the driving segment based on the vehicle time series data; the driving requirement is the driving data segment of the first specific SOC interval.

本发明相比于现有电池组一致性评估方法存在以下优点：Compared with the existing battery pack consistency evaluation method, the present invention has the following advantages:

(1)针对现有动力电池组一致性评估方法无法进行实车应用，或实车参数难以精确获取的问题，本发明所选取的开路电压、放电欧姆内阻、充电电压范数三个一致性表征参数可以通过车辆运行过程中BMS采集的信号计算得到，不需要精密的测量仪器或特定充放电工况，可在对电池无损耗的情况下在线检测电池一致性，应用场景广泛。(1) Aiming at the problem that the existing power battery pack consistency evaluation method cannot be used in real vehicles, or the real vehicle parameters are difficult to accurately obtain, the three consistency of the open circuit voltage, discharge ohmic resistance, and charging voltage norm selected by the present invention Characterization parameters can be calculated from the signals collected by the BMS during the operation of the vehicle. It does not require precise measuring instruments or specific charging and discharging conditions. The consistency of the battery can be checked online without battery loss. It has a wide range of application scenarios.

(2)针对现有一致性评估方法多基于单参数，难以全面、综合的反应电池组一致性情况的问题，本发明所提出的基于马氏距离和DBSCAN的一致性评估方法可以较好的评估不同数量级且具有相关性的多参数耦合动力电池一致性状态。(2) Aiming at the problem that the existing consistency evaluation methods are mostly based on a single parameter and it is difficult to comprehensively and comprehensively reflect the consistency of the battery pack, the consistency evaluation method based on Mahalanobis distance and DBSCAN proposed by the present invention can be better evaluated Coincidence state of multi-parameter coupling power battery with different orders of magnitude and correlation.

(3)针对现有在线参数辨识方法在开路电压辨识上常常出现的剧烈波动以及空值现象，本发明所提出的自适应OCV-RLS参数辨识方法可以有效的对异常OCV辨识值进行补偿，获得更准确的参数辨识结果，进而提高电池组一致性评估结果的准确性。(3) In view of the violent fluctuations and null phenomena that often occur in the open circuit voltage identification of the existing online parameter identification methods, the adaptive OCV-RLS parameter identification method proposed by the present invention can effectively compensate the abnormal OCV identification value to obtain More accurate parameter identification results, thereby improving the accuracy of the battery pack consistency evaluation results.

本说明书中各个实施例采用递进的方式描述，每个实施例重点说明的都是与其他实施例的不同之处，各个实施例之间相同相似部分互相参见即可。The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other.

本文中应用了具体个例对本发明的原理及实施方式进行了阐述，以上实施例的说明只是用于帮助理解本发明的方法及其核心思想；同时，对于本领域的一般技术人员，依据本发明的思想，在具体实施方式及应用范围上均会有改变之处。综上所述，本说明书内容不应理解为对本发明的限制。Specific examples are used in this article to illustrate the principles and implementation of the present invention. The description of the above examples is only used to help understand the method and core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the present invention There will be some changes in the specific implementation and application scope of the idea. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

一种电池组一致性评估方法，其特征在于，所述方法包括：A method for evaluating battery pack consistency, characterized in that the method includes:

步骤S1：对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段；Step S1: Divide the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment;

步骤S2：基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压；Step S2: Extract the open circuit voltage of the battery cell from the battery cell voltage based on the fully charged static segment;

步骤S3：基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻；Step S3: Based on the driving segment, establish an equivalent circuit model for the battery cells, use the adaptive OCV-RLS method to identify the parameters of the battery cells, and extract the ohmic internal resistance of each battery cell;

步骤S4：基于所述恒流充电片段，计算各电池单体的充电电压向量范数；Step S4: Calculate the charging voltage vector norm of each battery cell based on the constant current charging segment;

步骤S5：根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性；Step S5: According to the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, combining Mahalanobis distance and DBSCAN clustering algorithm to calculate the consistency of each battery pack under test at each evaluation point;

步骤S6：根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数；Step S6: Calculate the Z score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point;

步骤S7：根据Z分数确定待测电池组的一致性。Step S7: Determine the consistency of the battery pack to be tested according to the Z score.
根据权利要求1所述的电池组一致性评估方法，其特征在于，所述基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻，具体包括：The method for evaluating the consistency of a battery pack according to claim 1, wherein the equivalent circuit model is established for the battery cell based on the driving segment, and an adaptive OCV-RLS method is used for parameter identification of the battery cell , And extract the ohmic internal resistance of each battery cell, including:

步骤S31：确定待辨识参数估计公式；Step S31: Determine the parameter estimation formula to be identified;

步骤S32：采用差分电压方法确定初始时间步的电池开路电压估计值；Step S32: Use the differential voltage method to determine the estimated value of the battery open circuit voltage at the initial time step;

步骤S33：给定初始时间步的误差协方差矩阵和待辨识参数矩阵；Step S33: Given the error covariance matrix of the initial time step and the parameter matrix to be identified;

步骤S34：将第k-1时间步的误差协方差矩阵和待辨识参数矩阵代入所述待辨识参数估计公式，确定第k时间步的待辨识参数矩阵估计和误差协方差矩阵；Step S34: Substitute the error covariance matrix and the to-be-identified parameter matrix of the k-1 time step into the to-be-identified parameter estimation formula, and determine the to-be-identified parameter matrix estimation and the error covariance matrix of the k-th time step;

步骤S35：根据第k时间步的待辨识参数矩阵计算第k时间步的电池开路电压估计值；Step S35: Calculate the estimated value of the open circuit voltage of the battery at the k-th time step according to the to-be-identified parameter matrix at the k-th time step;

步骤S36：判断第k时间步的电池开路电压估计值是否满足电压异常判断规则；如果满足电压异常判断规则，则执行步骤S37；如果不满足电压异常判断规则，则执行步骤S38；所述电压异常判断规则为：定义电池放电电流为正，充电电流为负，当电池放电电流Ik为正时，电池开路电压估计值小于等于此刻的端电压，或大于前一时刻的电池开路电压估计值；若电池放电电流Ik为负时，电池开路电压估计值大于等于此刻的端电压，或小于前一时刻的电池开路电压估计值；Step S36: Determine whether the estimated value of the battery open circuit voltage at the k-th time step meets the voltage abnormality judgment rule; if the voltage abnormality judgment rule is met, go to step S37; if the voltage abnormality judgment rule is not met, go to step S38; the voltage is abnormal The judgment rule is: define the battery discharge current to be positive and the charge current to be negative. When the battery discharge current Ik is positive, the estimated value of the battery open circuit voltage is less than or equal to the terminal voltage at the moment, or greater than the estimated value of the battery open circuit voltage at the previous moment; if When the battery discharge current Ik is negative, the estimated value of the battery open circuit voltage is greater than or equal to the terminal voltage at the moment, or less than the estimated value of the battery open circuit voltage at the previous moment;

步骤S37：采用差分电压方法确定第k时间步的电池开路电压估计值，根据第k时间步的电池开路电压估计值更新待辨识参数矩阵估计值，并执行步骤S38；Step S37: Determine the estimated value of the open circuit voltage of the battery at the k-th time step by using the differential voltage method, update the estimated value of the parameter matrix to be identified according to the estimated value of the open-circuit voltage of the battery at the k-th time step, and execute step S38;

步骤S38：判断k是否小于或等于时间步总数n；如果k小于或等于n，则令k＝k+1，并返回步骤S34；如果k大于n，则将辨识得到的每个电池单体的欧姆内阻输出。Step S38: Judge whether k is less than or equal to the total number of time steps n; if k is less than or equal to n, set k=k+1, and return to step S34; if k is greater than n, the identification of each battery cell Ohm resistance output.
根据权利要求2所述的电池组一致性评估方法，其特征在于，所述采用差分电压方法确定第k时间步的电池开路电压估计值，具体包括：The method for evaluating battery pack consistency according to claim 2, wherein said determining the estimated value of the battery open circuit voltage at the k-th time step using a differential voltage method specifically comprises:

步骤S371：计算两个数据帧之间的电压差和电流差；Step S371: Calculate the voltage difference and the current difference between the two data frames;

步骤S372：以设定帧数据为一组，利用多个电压差和电流差构成的二维散点图，并对所述二维散点图进行线性回归分析，获得线性回归直线；Step S372: Taking the set frame data as a group, using a two-dimensional scatter diagram composed of multiple voltage differences and current differences, and performing linear regression analysis on the two-dimensional scatter diagram to obtain a linear regression line;

步骤S373：计算所述线性回归直线的斜率，并将所述斜率作为电池单体的内阻；Step S373: Calculate the slope of the linear regression line, and use the slope as the internal resistance of the battery cell;

步骤S374：基于所述电池单体的内阻计算电池开路电压估计值。Step S374: Calculate the estimated value of the open circuit voltage of the battery based on the internal resistance of the battery cell.
根据权利要求1所述的电池组一致性评估方法，其特征在于，所述根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性，具体包括：The method for evaluating the consistency of a battery pack according to claim 1, wherein the calculation of each battery cell is based on the open circuit voltage, ohmic resistance, and charging voltage vector norm of each battery cell, combined with the Mahalanobis distance and the DBSCAN clustering algorithm. The consistency of each battery pack to be tested at the evaluation point includes:

步骤S51：基于设定时间内所有车辆各电池单体对应的开路电压、欧姆内阻和充电电压向量范数构成第一评估参数矩阵；Step S51: Construct a first evaluation parameter matrix based on the open circuit voltage, ohmic internal resistance and charging voltage vector norm corresponding to each battery cell of all vehicles within the set time;

步骤S52：采用DBSCAN聚类算法，根据所述第一评估参数矩阵删除离群的电池单体，并获得第二评估参数矩阵；Step S52: Use the DBSCAN clustering algorithm to delete outlier battery cells according to the first evaluation parameter matrix, and obtain a second evaluation parameter matrix;

步骤S53：根据所述第二评估参数矩阵计算剩余电池单体的协方差矩阵；Step S53: Calculate the covariance matrix of the remaining battery cells according to the second evaluation parameter matrix;

步骤S54：在每一个评估点，根据待测车辆的当前评估参数矩阵，采用DBSCAN聚类算法删除离群的电池单体，并获得第三评估参数矩阵；Step S54: At each evaluation point, according to the current evaluation parameter matrix of the vehicle to be tested, use the DBSCAN clustering algorithm to delete outlier battery cells, and obtain a third evaluation parameter matrix;

步骤S55：根据所述第三评估参数矩阵计算在m个参数上的平均值，获得数据中心点；Step S55: Calculate the average value of the m parameters according to the third evaluation parameter matrix to obtain the data center point;

步骤S56：基于所述协方差矩阵与所述数据中心点确定各电池单体与数据中心点之间的马氏距离；Step S56: Determine the Mahalanobis distance between each battery cell and the data center point based on the covariance matrix and the data center point;

步骤S57：基于各电池单体与数据中心点之间的马氏距离计算各评估点处各待测电池组的一致性。Step S57: Calculate the consistency of each battery pack to be tested at each evaluation point based on the Mahalanobis distance between each battery cell and the data center point.
根据权利要求1所述的电池组一致性评估方法，其特征在于，所述对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段，具体包括：The method for evaluating the consistency of a battery pack according to claim 1, wherein the dividing the state of the vehicle to obtain a fully charged stationary segment, a driving segment, and a constant current charging segment specifically includes:

步骤S11：从大数据平台采集的车辆时间序列数据，所述车辆时间序列数据包括车辆速度、经纬度、SOC、电池组总电压、总电流和单体电压；Step S11: vehicle time series data collected from a big data platform, the vehicle time series data including vehicle speed, latitude and longitude, SOC, total battery voltage, total current and cell voltage;

步骤S12：基于所述车辆时间序列数据选取满足满电静置要求的连续时间序列片段作为满电静置片段；所述满电静置要求为连续1小时以上速度为零、电流为零，且SOC等于100％的数据点；Step S12: Based on the vehicle time series data, a continuous time sequence segment that meets the requirements of full-charge standing is selected as a fully-charged standing segment; the full-charge standing requirement is that the speed is zero and the current is zero for more than 1 hour, and Data points where SOC is equal to 100%;

步骤S13：基于所述车辆时间序列数据选取满足行驶要求的连续时间序列片段作为行驶片段；所述行驶要求为第一特定SOC区间的行驶数据片段；Step S13: Based on the vehicle time series data, a continuous time sequence segment that meets the driving requirement is selected as a driving segment; the driving requirement is a driving data segment of a first specific SOC interval;

步骤S14：基于所述车辆时间序列数据选取满足恒流充电要求的连续时间序列片段作为恒流充电片段；所述恒流充电要求为在第二特定SOC区间内充电电流为恒定值。Step S14: Based on the vehicle time series data, a continuous time series segment that meets the constant current charging requirement is selected as the constant current charging segment; the constant current charging requirement is that the charging current is a constant value in the second specific SOC interval.
一种电池组一致性评估***，其特征在于，所述***包括：A battery pack consistency evaluation system, characterized in that the system includes:

状态划分模块，用于对车辆状态进行划分，获得满电静置片段、行驶片段以及恒流充电片段；The state division module is used to divide the state of the vehicle to obtain the fully charged stationary segment, the driving segment and the constant current charging segment;

开路电压确定模块，用于基于所述满电静置片段，通过电池单体电压提取电池单体的开路电压；The open circuit voltage determination module is configured to extract the open circuit voltage of the battery cell from the battery cell voltage based on the fully charged static segment;

欧姆内阻确定模块，用于基于所述行驶片段，针对电池单体建立等效电路模型，采用自适应OCV-RLS方法对电池单体进行参数辨识，并提取每个电池单体的欧姆内阻；The ohmic resistance determination module is used to establish an equivalent circuit model for the battery cell based on the driving segment, use the adaptive OCV-RLS method to identify the battery cell parameters, and extract the ohmic resistance of each battery cell ；

范数确定模块，用于基于所述恒流充电片段，计算各电池单体的充电电压向量范数；A norm determining module, configured to calculate the norm of the charging voltage vector of each battery cell based on the constant current charging segment;

第一一致性确定模块，用于根据各电池单体的开路电压、欧姆内阻和充电电压向量范数，结合马氏距离与DBSCAN聚类算法计算各评估点处各待测电池组的一致性；The first consistency determination module is used to calculate the consistency of each battery pack under test at each evaluation point based on the open circuit voltage, ohmic resistance and charging voltage vector norm of each battery cell, combined with Mahalanobis distance and DBSCAN clustering algorithm sex;

Z分数确定模块，用于根据各评估点处所有待测电池组的一致性计算各评估点处待测电池组的Z分数；The Z-score determination module is used to calculate the Z-score of the battery pack to be tested at each evaluation point according to the consistency of all battery packs to be tested at each evaluation point;

第二一致性确定模块，用于根据Z分数确定待测电池组的一致性。The second consistency determination module is used to determine the consistency of the battery pack to be tested according to the Z score.
根据权利要求6所述的电池组一致性评估***，其特征在于，所述欧姆内阻确定模块，具体包括：The battery pack consistency evaluation system according to claim 6, wherein the ohmic internal resistance determination module specifically comprises:

待辨识参数估计公式确定单元，用于确定待辨识参数估计公式；The parameter estimation formula determination unit to be identified is used to determine the parameter estimation formula to be identified;

初始时间步的电池开路电压估计值确定单元，用于采用差分电压方法确定初始时间步的电池开路电压估计值；The battery open circuit voltage estimation value determination unit at the initial time step is used to determine the battery open circuit voltage estimation value at the initial time step by using a differential voltage method;

给定单元，用于给定初始时间步的误差协方差矩阵和待辨识参数矩阵；A given unit is used for a given initial time step error covariance matrix and parameter matrix to be identified;

参数确定单元，用于将第k-1时间步的误差协方差矩阵和待辨识参数矩阵代入所述待辨识参数估计公式，确定第k时间步的待辨识参数矩阵估计和误差协方差矩阵；The parameter determination unit is configured to substitute the error covariance matrix and the parameter matrix to be identified at the k-1 time step into the parameter estimation formula to be identified, and determine the parameter matrix estimation and error covariance matrix to be identified at the k time step;

第k时间步的电池开路电压估计值确定单元，用于根据第k时间步的待辨识参数矩阵计算第k时间步的电池开路电压估计值；The battery open-circuit voltage estimated value determination unit at the k-th time step is configured to calculate the battery open-circuit voltage estimated value at the k-th time step according to the to-be-identified parameter matrix at the k-th time step;

判断单元，用于判断第k时间步的电池开路电压估计值是否满足电压异常判断规则；如果满足电压异常判断规则，则执行“更新单元”；如果不满足电压异常判断规则，则执行“第二判断单元”；所述电压异常判断规则为：定义电池放电电流为正，充电电流为负，当电池放电电流Ik为正时，电池开路电压估计值小于等于此刻的端电压，或大于前一时刻的电池开路电压估计值；若电池放电电流Ik为负时，电池开路电压估计值大于等于此刻的端电压，或小于前一时刻的电池开路电压估计值；The judgment unit is used to judge whether the estimated value of the battery open circuit voltage at the kth time step meets the voltage abnormality judgment rule; if it meets the voltage abnormality judgment rule, execute the "update unit"; if it does not meet the voltage abnormality judgment rule, execute the "second Judgment unit"; The voltage abnormality judgment rule is: define the battery discharge current as positive and the charging current as negative. When the battery discharge current Ik is positive, the estimated value of the battery open circuit voltage is less than or equal to the terminal voltage at this moment, or greater than the previous moment If the battery discharge current Ik is negative, the battery open circuit voltage estimate is greater than or equal to the terminal voltage at the moment, or less than the battery open circuit voltage estimate at the previous moment;

更新单元，用于采用差分电压方法确定第k时间步的电池开路电压估计值，根据第k时间步的电池开路电压估计值更新待辨识参数矩阵估计值，并执行“第二判断单元”；The update unit is used to determine the estimated value of the open circuit voltage of the battery at the k-th time step by using the differential voltage method, update the estimated value of the parameter matrix to be identified according to the estimated value of the open-circuit voltage of the battery at the k-th time step, and execute the "second judgment unit";

第二判断单元，用于判断k是否小于或等于时间步总数n；如果k小于或等于n，则令k＝k+1，并返回“参数确定单元”；如果k大于n，则将辨识得到的每个电池单体的欧姆内阻输出。The second judgment unit is used to judge whether k is less than or equal to the total number of time steps n; if k is less than or equal to n, set k=k+1 and return to the "parameter determination unit"; if k is greater than n, it will be identified The ohmic internal resistance output of each battery cell.
根据权利要求7所述的电池组一致性评估***，其特征在于，所述更新单元，具体包括：The battery pack consistency evaluation system according to claim 7, wherein the updating unit specifically comprises:

电压差和电流差确定子单元，用于计算两个数据帧之间的电压差和电流差；The voltage difference and current difference determining subunit is used to calculate the voltage difference and current difference between two data frames;

线性回归分析子单元，用于以设定帧数据为一组，利用多个电压差和电流差构成的二维散点图，并对所述二维散点图进行线性回归分析，获得线性回归直线；The linear regression analysis subunit is used to set the frame data as a group, use a two-dimensional scatter diagram formed by multiple voltage differences and current differences, and perform linear regression analysis on the two-dimensional scatter diagram to obtain a linear regression straight line;

斜率计算确定子单元，用于计算所述线性回归直线的斜率，并将所述斜率作为电池单体的内阻；A slope calculation and determination subunit, used to calculate the slope of the linear regression line, and use the slope as the internal resistance of the battery cell;

电池开路电压估计值确定子单元，用于基于所述电池单体的内阻计算电池开路电压估计值；A battery open circuit voltage estimation value determination subunit, which is used to calculate the battery open circuit voltage estimation value based on the internal resistance of the battery cell;

更新子单元，用于根据电池开路电压估计值更新待辨识参数矩阵估计值。The update subunit is used to update the estimated value of the parameter matrix to be identified according to the estimated value of the open circuit voltage of the battery.
根据权利要求6所述的电池组一致性评估***，其特征在于，所述第一一致性确定模块，具体包括：The battery pack consistency evaluation system according to claim 6, wherein the first consistency determination module specifically comprises:

第一评估参数矩阵确定单元，用于基于设定时间内所有车辆各电池单体对应的开路电压、欧姆内阻和充电电压向量范数构成第一评估参数矩阵；The first evaluation parameter matrix determination unit is configured to form a first evaluation parameter matrix based on the open circuit voltage, ohmic internal resistance and charging voltage vector norm corresponding to each battery cell of all vehicles within a set time;

第二评估参数矩阵确定单元，用于采用DBSCAN聚类算法，根据所述第一评估参数矩阵删除离群的电池单体，并获得第二评估参数矩阵；The second evaluation parameter matrix determination unit is configured to adopt the DBSCAN clustering algorithm to delete outlier battery cells according to the first evaluation parameter matrix, and obtain a second evaluation parameter matrix;

协方差矩阵确定单元，用于根据所述第二评估参数矩阵计算剩余电池单体的协方差矩阵；A covariance matrix determining unit, configured to calculate the covariance matrix of the remaining battery cells according to the second evaluation parameter matrix;

第三评估参数矩阵确定单元，用于在每一个评估点，根据待测车辆的当前评估参数矩阵，采用DBSCAN聚类算法删除离群的电池单体，并获得第三评估参数矩阵；The third evaluation parameter matrix determination unit is used to use the DBSCAN clustering algorithm to delete outlier battery cells at each evaluation point according to the current evaluation parameter matrix of the vehicle to be tested, and obtain the third evaluation parameter matrix;

数据中心点确定单元，用于根据所述第三评估参数矩阵计算在m个参数上的平均值，获得数据中心点；A data center point determining unit, configured to calculate an average value on m parameters according to the third evaluation parameter matrix to obtain a data center point;

马氏距离确定单元，用于基于所述协方差矩阵与所述数据中心点确定各电池单体与数据中心点之间的马氏距离；A Mahalanobis distance determining unit, configured to determine the Mahalanobis distance between each battery cell and the data center point based on the covariance matrix and the data center point;

第一一致性确定单元，用于基于各电池单体与数据中心点之间的马氏距离计算各评估点处各待测电池组的一致性。The first consistency determining unit is used to calculate the consistency of each battery pack to be tested at each evaluation point based on the Mahalanobis distance between each battery cell and the data center point.
根据权利要求6所述的电池组一致性评估***，其特征在于，所述状态划分模块，具体包括：The battery pack consistency evaluation system according to claim 6, wherein the state division module specifically comprises:

车辆时间序列数据获取单元，用于从大数据平台采集的车辆时间序列数据，所述车辆时间序列数据包括车辆速度、经纬度、SOC、电池组总电压、总电流和单体电压；A vehicle time series data acquisition unit for vehicle time series data collected from a big data platform, the vehicle time series data including vehicle speed, latitude and longitude, SOC, total battery pack voltage, total current, and cell voltage;

满电静置片段确定单元，用于基于所述车辆时间序列数据选取满足满电静置要求的连续时间序列片段作为满电静置片段；所述满电静置要求为连续1小时以上速度为零、电流为零，且SOC等于100％的数据点；A fully charged stationary segment determining unit, configured to select a continuous time sequence segment that meets the requirements of fully charged stationary as a fully charged stationary segment based on the vehicle time series data; the fully charged stationary requirement is that the speed of the fully charged stationary is more than 1 hour continuously. Zero, the current is zero, and the SOC is equal to the data point of 100%;

行驶片段确定单元，用于基于所述车辆时间序列数据选取满足行驶要求的连续时间序列片段作为行驶片段；所述行驶要求为第一特定SOC区间的行驶数据片段；A driving segment determination unit, configured to select a continuous time sequence segment that meets a driving requirement as a driving segment based on the vehicle time series data; the driving requirement is a driving data segment of a first specific SOC interval;

恒流充电片段确定单元，用于基于所述车辆时间序列数据选取满足恒流充电要求的连续时间序列片段作为恒流充电片段；所述恒流充电要求为在第二特定SOC区间内充电电流为恒定值。The constant current charging segment determination unit is configured to select a continuous time series segment that meets the constant current charging requirement as the constant current charging segment based on the vehicle time series data; the constant current charging requirement is that the charging current in the second specific SOC interval is Constant value.