CN114879049A - Power battery consistency safety state evaluation method - Google Patents

Abstract

The invention relates to the technical field of power battery evaluation, in particular to a power battery consistency safety state evaluation method, which comprises the following steps: extracting a plurality of charging data segments capable of reflecting the vehicle state; calculating the standard deviation characteristic and the variance entropy consistency characteristic of the monomer voltage of each charging data segment to obtain a characteristic value; acquiring the covering times of a charging process of covering a target preset interval in the full life cycle of the vehicle, and combining the covering times with the first correction times and the second correction times to respectively obtain a reference sample and an evaluation sample; constructing the characteristic values into a characteristic matrix, and performing unsupervised training after processing to divide the characteristic values into two types to obtain a confusion matrix; constructing a power battery consistency safety state quantitative calculation model; and constructing a state evaluation alarm grade model, and outputting a grade result to represent the safety state of the power battery. The method and the system can timely warn the potential risks of the vehicle and avoid the accident risk caused by the abnormal state evolution of the vehicle.

Description

Evaluation method of power battery consistency safety state

technical field

The invention relates to the technical field of power battery evaluation, in particular to a power battery consistency safety state evaluation method.

Background technique

Compared with existing fuel-powered vehicles, new energy vehicles have a short development history. The development of new energy vehicle technologies is not as mature as that of fuel-powered vehicles. The exploration and research of new energy vehicles and their power batteries is far less than that of fuel-powered vehicles. thorough. However, the actual operation of new energy vehicles has the characteristics of changeable environment, complex scene, multi-dimensional, redundant, heterogeneous and strong coupling of data, which brings great challenges to the exploration and mining of the safety status of power vehicles contained in the operation of big data, which makes the strong coupling complex. The systematic evaluation of the power battery status of new energy vehicles is difficult to achieve through mechanism analysis.

Since the new energy vehicle will upload the operating data according to the standard GB-32960 during the service period, and the characteristic signal such as voltage is a kind of characteristic quantity of the comprehensive state of the power battery, so the safety features of the historical data of the power vehicle operation can be analyzed and mined. Further quantify the degree of difference in the state of the power battery during service.

Therefore, in order to discover the safety problems of power vehicles in time, an electrochemical mechanism that can not consider complex and coupled electrochemical mechanisms is urgently needed, and the overall signal safety degree of the power battery system can be described directly by calculating the data characteristics of the signal to realize the battery system. The quantification of the safety state of the system realizes the rapid quantification and evaluation of the safety state of the battery, and then guides the power vehicle to carry out safety warning.

SUMMARY OF THE INVENTION

The present invention intends to provide a power battery consistency safety state evaluation method to quickly quantify and evaluate the power battery safety state.

The power battery consistency safety state assessment method in this solution includes the following steps:

Step 1: Obtain historical operating data of the vehicle to be evaluated, and extract a plurality of charging data segments that can reflect the state of the vehicle from the historical operating data;

Step 2: Calculate the standard deviation feature and the variance entropy consistency feature for the cell voltage of each charging data segment to obtain the feature value;

Step 3: Obtain the coverage times of the charging process covering the target preset interval in the entire life cycle of the vehicle, multiply the coverage times by the first parameter to obtain the first correction times, and use the feature value of the first correction times in the charging data segment as a reference sample. , multiply the coverage times by the second parameter to obtain the second correction times, and take the feature value of the second correction times in the charging data segment as the evaluation sample;

Step 4: Construct the eigenvalues into a eigenmatrix, delete the eigenvalues greater than the 99th percentile in the eigenvalues, perform unsupervised training of a preset algorithm on the eigenvalues, and divide them into two categories to obtain a confusion matrix;

Step 5: Construct a power battery consistent safety state quantitative calculation model according to the confusion matrix, and obtain a consistent state safety assessment score by calculating the consistent safety state quantitative calculation model;

In step 6, a state evaluation alarm level model is constructed according to the consistency state safety evaluation score, and the level result is output, and the safety state of the power battery is represented by the level result.

The beneficial effects of this program are:

By slicing the charging data in the historical operation data of the new energy vehicle, the charging data segment is obtained, the charging feature extraction is completed, and then the evaluation score is obtained by combining the consistency state safety evaluation method, and the state evaluation alarm level is finally output to warn the potential risks of the vehicle in time. , to avoid the abnormal state of the vehicle from evolving into a more serious accident risk.

Further, in the first step, the battery OCV curve of the vehicle to be evaluated is obtained, and charging data segments are extracted from the SOC interval on the OCV curve.

The beneficial effect is that the safety state of the vehicle can be accurately reflected on the premise of retaining the corresponding features.

Further, in the second step, the calculation formula of the standard deviation feature is:

The formula for calculating the variance entropy consistency feature is:

The beneficial effects are: through the calculation of the standard deviation feature, the discrete degree of the data can be quickly and accurately analyzed, and through the definition of the variance entropy used to describe the degree of disorder of the system, the voltage difference between the individual cells of the power battery system can be described, that is, The consistency is good or bad, the smaller the variance entropy, the smaller the cell voltage difference, the better the consistency, and the safer the battery system.

Further, in the third step, the number of coverages is denoted as n, the first parameter is denoted as α, the second parameter is denoted as β, the first number of corrections is n*α, and the second number of corrections is n*β.

The beneficial effect is that the calculation of the quantization process is smoother by representing each data in characters.

Further, the value range of the first parameter α is: (0.05, 0.5), and the value range of the second parameter is: (0.01, 0.2).

The beneficial effects are: by limiting the value range of the first parameter and the second parameter, the number of features in the data set can be reduced, and the amount of calculation can be reduced, and the first parameter and the second parameter can be valued in different ranges, which can increase the number of features in the data set. Differences between a large reference sample and evaluation data.

Further, in the fourth step, the feature matrix is represented as X, and the confusion matrix obtained after unsupervised training on the feature matrix is:

The beneficial effects are: by calculating the confusion matrix, the feature matrix can be simply classified, and the processing speed of the historical operation data can be improved.

Further, in the step 5, the consistent security state quantitative calculation model is:

表示参考样本中方差特征的均值，

表示参考样本中方差熵一致性特征的均值，

表示评估样本中的方差特征均值，

表示评估样本中的方差熵一致性特征均值，μ和

表示权重，权重的取值范围为[0,1]。Among them, S represents the consistency state security evaluation score of the evaluation sample, #(X _{0_1} ), #(X ₀ ), #(X _{1_1} ), #(X ₁ ) represent (X _{0_1} ), X ₀ , X respectively _{1_1} , the number of X ₁ samples,

represents the mean of variance features in the reference sample,

represents the mean of variance entropy consistency features in the reference sample,

represents the mean of variance features in the evaluation sample,

Represents the variance entropy consistency feature mean in the evaluation sample, μ and

Indicates the weight, and the value range of the weight is [0,1].

The beneficial effects are as follows: through the quantification of the number of samples, the degree of consistency and security between the reference data and the data to be evaluated can be quantified; The size of the degree can be quantified to obtain the safety assessment score and improve the accuracy of the assessment.

Further, in the step 6, the state assessment alarm level model is:

Among them, alarmLevel represents the alarm level obtained from the state evaluation of the data to be evaluated. The alarm levels 0, 1, 2 and 3 correspond to no, low, medium and high risks respectively, r=1-S represents the vehicle risk, r ₁ , r ₂ , r ₃ corresponds to the risk alarm threshold respectively, and r ₁ <r ₂ <r ₃ .

The beneficial effects are: by quantitatively evaluating the consistent safety state of the power battery, risk factors can be found more intuitively from a large amount of historical operation data analysis.

Description of drawings

FIG. 1 is a flowchart of an embodiment of a power battery consistency safety state assessment method according to the present invention;

2 is an OCV curve diagram in an embodiment of a power battery consistency safety state evaluation method according to the present invention;

FIG. 3 is a voltage data diagram of a single charging data segment in an embodiment of a power battery consistency safety state evaluation method according to the present invention.

Detailed ways

The following is further described in detail through specific embodiments.

Example

The power battery consistency safety state assessment method, as shown in Figure 1, includes the following steps:

Step 1: Obtain the historical operating data of the vehicle to be evaluated, and obtain the battery OCV curve of the vehicle to be evaluated. The OCV curve is a reference tool for power batteries. Each battery is given by the manufacturer when it is listed and extracted from the historical operating data. Multiple charging data segments that can reflect the state of the vehicle. For example, taking 100 charging data segments as an example, the voltage data of a single charging data segment is shown in Figure 3, and the preset time period when the vehicle first starts to serve is defined as the vehicle safety state. The reference state, that is, the new cars are all good by default, extract the charging data segment from the SOC interval on the OCV curve, that is, the extracted charging data segment is the charging data containing the charge from 85% to 95%, and select the SOC interval according to the OCV curve. The purpose is: in this SOC interval, there is a sudden adjustment of the voltage, which is equivalent to giving a typical excitation, and then the signal change based on this excitation can represent the battery state, as shown in the box in Figure 2.

Step 2: Calculate the standard deviation feature and the variance entropy consistency feature for the cell voltage of each charging data segment, and obtain two eigenvalues. Then, after 100 charging data segments are calculated, a 100×2 eigenvalue matrix can be obtained. The voltage is the voltage value of the single cell, the single voltage is expressed as C, and the calculation formula of the standard deviation characteristic is:

其中，x_ij为第i时刻第j个电芯的单体电压值，x_i为i时刻所有单体电压的平均值；

Among them, x _ij is the cell voltage value of the j-th cell at the i-th time, and x _i is the average value of all cell voltages at the i-th time;

The formula for calculating the variance entropy consistency feature is:

其中，E_j为单体电压值的平方。

Among them, E _j is the square of the cell voltage value.

Step 3: Obtain the coverage times of the charging process covering the target preset interval in the whole life cycle of the vehicle, the coverage times are represented as n, and the coverage times are multiplied by the first parameter to obtain the first correction times, the first parameter is represented as α, the first The value range of parameter α is: (0.05, 0.5), the first correction times is n*α, and the feature value of the first correction times in the charging data segment is used as a reference sample, and the coverage times are multiplied by the second parameter to obtain the first number of corrections. Second correction times, the second parameter is represented as β, the value range of the second parameter is: (0.01, 0.2), the second correction times is n*β, and the feature value of the second correction times in the charging data segment is used as Evaluate samples.

In step 3, "before" means that the acquired charging data segments are acquired according to the starting point of time, and "after" means that the acquired charging data segments are acquired from the end of time. For example, the acquired charging data segments are 100 The first correction times can be set to 20%, the second correction times can be set to 10%, and 100 pieces of data are taken from the data fragments of the first correction times before. The first 20 pieces of data in the clips were used as the control group, and the last 10 data clips of the 100 data clips were taken as the evaluation samples.

Step 4: Construct the eigenvalues into a eigenmatrix, denoted as X, delete the eigenvalues greater than the 99th percentile in the eigenvalues, and perform unsupervised training of the preset algorithm on the eigenmatrix, and the preset algorithm is the existing K- The Means algorithm divides X into two categories, the 0th category and the 1st category, respectively, and the confusion matrix is obtained, which is:

其中X₀为参考样本，X₁为评估样本。

Among them, X ₀ is the reference sample, and X ₁ is the evaluation sample.

Among them, X _{0_0} means dividing the 0 group of the reference sample into the 0th class, X _{0_1} means dividing the 0 group of the reference sample into the 1st class, X _{1_0} means dividing the 1 group of the evaluation sample into the 0th class, X _{1_1} means Group 1 of the evaluation samples into category 1.

Step 5: Based on the reference samples and evaluation samples in the confusion matrix, construct a quantitative calculation model of the consistent safety state of the power battery according to the confusion matrix, and calculate the consistent state safety evaluation score through the consistent safety state quantitative calculation model. The state quantitative calculation model is:

表示参考样本中方差特征的均值，

表示参考样本中方差熵一致性特征的均值，

表示评估样本中的方差特征均值，

表示评估样本中的方差熵一致性特征均值，μ和

表示权重，权重的取值范围为[0,1]。Among them, S represents the consistency state security evaluation score of the evaluation sample, #(X _{0_1} ), #(X ₀ ), #(X _{1_1} ), #(X ₁ ) represent (X _{0_1} ), X ₀ , X respectively _{1_1} , the number of X ₁ samples,

represents the mean of variance features in the reference sample,

represents the mean of variance entropy consistency features in the reference sample,

represents the mean of variance features in the evaluation sample,

Represents the variance entropy consistency feature mean in the evaluation sample, μ and

Indicates the weight, and the value range of the weight is [0,1].

At the beginning, the state of the vehicle battery is good, which is equivalent to taking the two eigenvalues of the reference sample as the initial origin, and the closer the two eigenvalues calculated by the evaluation sample are to the initial origin, the closer the state of the surface vehicle battery is to the initial state. , the better the safety of the vehicle. With the use of the vehicle, the battery of the vehicle will get worse and worse, and the data in the subsequent use of the vehicle is used as the evaluation sample for classification, and then the corresponding score is calculated. That is, the higher the proportion of the evaluation sample group 1 classified as the 0th group represented by (X _{0_1} ), the closer the evaluation sample is to the reference sample, the smaller the difference between the evaluation sample and the reference sample, the smaller the separation degree, and the safety of the vehicle. the higher.

Step 6: Build a state evaluation alarm level model according to the consistency state safety evaluation score, that is, the vehicle risk is obtained by subtracting the consistency state safety evaluation score from the total number, and the level result is output, that is, the vehicle risk is expressed as r=1-S, and the level is expressed as r=1-S. The results represent the safe state of the power battery, and the state evaluation alarm level model is:

Among them, alarmLevel represents the alarm level obtained from the state evaluation of the data to be evaluated. The alarm levels 0, 1, 2 and 3 correspond to no, low, medium and high risks respectively, r=1-S represents the vehicle risk, r ₁ , r ₂ , r ₃ corresponds to the risk alarm threshold respectively, and r ₁ <r ₂ <r ₃ , the risk alarm threshold is set according to actual needs, such as 40, 60, and 80.

In this embodiment, by slicing the historical charging data of the new energy vehicle, the charging data segment is obtained, the charging feature extraction is completed, and then the evaluation score is obtained in combination with the consistency state safety evaluation method, and the state evaluation alarm level is finally output, so as to timely warn the potential risks of the vehicle. Avoid the abnormal state of the vehicle from evolving into a more serious accident risk.

The above descriptions are only embodiments of the present invention, and common knowledge such as well-known specific structures and characteristics in the solution are not described too much here. It should be pointed out that for those skilled in the art, some modifications and improvements can be made without departing from the structure of the present invention. These should also be regarded as the protection scope of the present invention, and these will not affect the implementation of the present invention. Effectiveness and utility of patents. The scope of protection claimed in this application shall be based on the content of the claims, and the descriptions of the specific implementation manners in the description can be used to interpret the content of the claims.

Claims (8)

1. The method for evaluating the consistency safety state of the power battery is characterized by comprising the following steps of:

acquiring historical operation data of a vehicle to be evaluated, and extracting a plurality of charging data segments capable of reflecting the vehicle state from the historical operation data;

calculating the standard deviation characteristic and the variance entropy consistency characteristic of the monomer voltage of each charging data segment to obtain a characteristic value;

step three, acquiring the covering times of the charging process of the vehicle in the full life cycle covering target preset interval, multiplying the covering times by a first parameter to obtain a first correction time, taking the characteristic value of the first correction time in the charging data segment as a reference sample, multiplying the covering times by a second parameter to obtain a second correction time, and taking the characteristic value of the second correction time in the charging data segment as an evaluation sample;

constructing the characteristic values into a characteristic matrix, deleting the characteristic values of which the quantiles are more than 99 quantiles in the characteristic matrix, carrying out unsupervised training of a preset algorithm on the characteristic matrix, and dividing the characteristic matrix into two types to obtain a confusion matrix;

constructing a power battery consistency safety state quantitative calculation model according to the confusion matrix, and calculating to obtain a consistency state safety evaluation score through the consistency safety state quantitative calculation model;

and step six, constructing a state evaluation alarm grade model according to the consistency state safety evaluation score, outputting a grade result, and representing the safety state of the power battery by using the grade result.

2. The power battery consistency safety state evaluation method according to claim 1, characterized in that: in the first step, an OCV curve of a battery of the vehicle to be evaluated is obtained, and a charging data segment is extracted from an SOC interval on the OCV curve.

3. The power battery consistency safety state evaluation method according to claim 2, characterized in that: in the second step, the calculation formula of the standard deviation features is as follows:

the calculation formula of the variance entropy consistency characteristic is as follows:

4. the power battery consistency safety state evaluation method according to claim 3, characterized in that: in the third step, the number of covering times is represented as n, the first parameter is represented as α, the second parameter is represented as β, the first correction time is n × α, and the second correction time is n × β.

5. The power battery consistency safety state evaluation method according to claim 4, characterized in that: the value range of the first parameter alpha is as follows: (0.05, 0.5), wherein the value range of the second parameter is as follows: (0.01,0.2).

6. The power battery consistency safety state evaluation method according to claim 5, characterized in that: in the fourth step, the feature matrix is represented as X, and the confusion matrix obtained after the feature matrix is subjected to unsupervised training is as follows:

7. the power battery consistency safety state evaluation method according to claim 5, characterized in that: in the fifth step, the consistency safety state quantitative calculation model is as follows:

wherein S represents a consistency state safety assessment score, # (X) of the assessment sample _{0_1} )、#(X ₀ )、#(X _{1_1} )、#(X ₁ ) Respectively represent (X) _{0_1} )、X ₀ 、X _{1_1} 、X ₁ The number of the samples is such that,

representing the mean of the variance features in the reference sample,

represents the mean of the variance entropy consistency feature in the reference sample,

representing the mean of the variance features in the evaluation sample,

means for variance entropy consistency feature means, μ sum, in the evaluation samples

Represents weight, and the value range of the weight is [0,1 ]]。

8. The power battery consistency safety state evaluation method according to claim 5, characterized in that: in the sixth step, the state assessment alarm grade model is as follows:

the alarmLevel represents the alarm level obtained by evaluating the state of data to be evaluated, 0-level, 1-level, 2-level and 3-level alarms respectively correspond to no risk, low risk, medium risk and high risk, r is 1-S represents the vehicle risk, and r is ₁ 、r ₂ 、r ₃ Respectively corresponding to a risk alarm threshold, and r ₁ <r ₂ <r ₃ 。

Images