A method for enhancing the stability of battery health diagnosis results

Abstract

The present invention discloses a method for enhancing the stability of battery health diagnosis results, which relates to the field of new energy battery application technology. The present invention first collects the SOH estimation results SOH(i) of the battery system within a set time period, as well as the corresponding date D(i) and the lowest SOC value SOC of the battery data used on that day. end (i), then by introducing SOC avg , dE, and λ, and devises probability density distribution calculation methods with varying degrees of confidence to assist in assessing the reliability of each SOH(i) data point, ultimately determining the accuracy and reliability of the current SOH estimate. The algorithm provided by this invention addresses the issue of unstable battery system capacity estimation results in existing solutions, improving the accuracy of SOH estimates.

Description

Technical Field

[0001] The present invention relates to the field of new energy battery application technology, and in particular to a method for enhancing the stability of battery health diagnosis results. Background Art

[0002] After a power battery system is installed and used on a vehicle, its available capacity will decay over time. Due to varying operating environments and conditions, the rate of capacity decay varies from vehicle to vehicle. This capacity decay is typically caused by excessive capacity decay or abnormal self-discharge rates in individual cells. The available capacity of a battery system determines the vehicle's mileage. Accurately assessing the available capacity accurately reflects the health of the vehicle's battery system, is a prerequisite for accurate mileage estimation, and is an effective preventative measure to prevent vehicle breakdowns.

[0003] The battery system's state of health (SOH) is the primary key health indicator used to assess battery health. While many SOH estimation methods exist, their accuracy is limited, with SOH errors typically reaching 8% or more. Therefore, high-precision SOH estimation has long been an industry challenge. Existing methods for estimating battery system SOH use battery voltage and current data to estimate the battery system capacity by building a battery model. However, the estimated results are susceptible to data quality, battery depth of discharge, vehicle operating conditions, and other factors, resulting in unstable results such as fluctuations over time. Summary of the Invention

[0004] The present invention provides a method for enhancing the stability of battery health diagnosis results, the main purpose of which is to solve the problem that the SOH estimation method in the prior art is easily affected by various factors and causes unstable estimation results.

[0005] The present invention adopts the following technical solutions:

[0006] A method for enhancing the stability of battery health diagnosis results includes the following steps:

[0007] Step S1: Obtain the number of SOH estimations of the battery system within a set time period n, record the SOH estimation results SOH(i) of each time, the corresponding date D(i) and the lowest SOC value SOC of the battery data used on that day end (i);

[0008] Step S2: Introducing SOC avg , dE and λ are three evaluation parameters, among which SOC avg Indicates each of the SOC end(i), dE represents the stability of each SOH (i), and λ represents the sum of the reliability scores of each SOH estimation result depending on the minimum SOC value. The values ​​of each evaluation parameter are calculated respectively by formulas (1)-(4):

[0009]

[0010] dE=2×[max(SOH(i))-mean(SOH(i))] (2)

[0011]

[0012]

[0013] Where: α(i) is the reliability score of the i-th SOH estimation result depending on the minimum SOC value;

[0014] Step S3: SOC avg The three evaluation parameters , dE, and λ are used to set the membership function respectively, thus forming the membership matrix A. The membership matrix A is a 3*3 matrix, and the three columns represent the three credibility types of SOH results: untrustworthy, generally credible, and very credible; the three rows represent SOC avg , dE, and λ are the membership degrees of each credibility corresponding to the three evaluation parameters;

[0015] Step S4: SOC avg The weight coefficients of the three evaluation parameters , dE, and λ are set, and the probability density distribution matrix P of each credibility is calculated according to formula (5): j :

[0016]

[0017] Where: ω k For SOC avg , dE, λ are the weight coefficients of the three parameters; k is the row number of matrix A; j is the column number of matrix A;

[0018] Step S5: Matrix P j The credibility category of the position corresponding to the maximum probability density value is determined as the credibility classification calculation result of the current SOH estimation value SOH(n), thereby confirming the stability and accuracy of the current SOH estimation value SOH(n).

[0019] Further, in step S5, the credibility K is used to represent the matrix P jThe credibility classification calculation result is that K=1 represents untrustworthy, K=2 represents generally trustworthy, and K=3 represents very trustworthy. By comparing the credibility K obtained in this evaluation with the credibility K0 obtained in the previous evaluation, the battery system SOH1 after this evaluation is obtained.

[0020] Furthermore, if K>K0, or K=K0 and the duration exceeds the set value, the battery system SOH1 after this evaluation is equal to the current SOH estimation value SOH(n); if K<K0, the battery system SOH1 after this evaluation is equal to the battery system SOH0 after the previous evaluation.

[0021] Furthermore, in step S1 , SOH(i) is the i-th SOH value of the available capacity of the battery system or the SOH value of the available capacity of a single cell in the battery system.

[0022] Furthermore, in step S1, the time period is set to a specified time length or a time period corresponding to when the estimated frequency reaches a specified number of times.

[0023] Furthermore, in step S2, the membership function may be a linear trigonometric function, a rectangular function or a Gaussian function.

[0024] Compared with the prior art, the present invention has the following beneficial effects:

[0025] 1. The present invention introduces SOC avg , dE, and λ, and designs probability density distribution value calculation methods with different credibility to assist in evaluating whether the data of each SOH(i) is reliable, and then finally determine whether the current SOH estimation value is accurate and reliable, thereby effectively improving the stability of the SOH diagnosis results and contributing to the accurate evaluation of the health status of the vehicle battery system.

[0026] 2. The algorithm provided by the present invention is flexible and can adapt to different scenarios by adjusting the membership function or the weight coefficients of different credibility modes. It has strong versatility and a short calculation process, and is suitable for online estimation of battery health diagnosis. DETAILED DESCRIPTION

[0027] The specific embodiments of the present invention are described below. In order to fully understand the present invention, many details are described below, but for those skilled in the art, the present invention can be implemented without these details.

[0028] A method for enhancing the stability of battery health diagnosis results includes the following steps:

[0029] Step S1: Obtain the number of SOH estimations of the battery system within a set time period n, record the SOH estimation results SOH(i) of each time, the corresponding date D(i) and the lowest SOC value SOC of the battery data used on that dayend (i).

[0030] In this step, SOH(i) is the i-th SOH value of the battery system's available capacity or the SOH value of the available capacity of a single cell in the battery system. The time period is set to a specified length of time (e.g., 4 months) or the time period corresponding to when the estimation frequency reaches a specified number of times (e.g., 10 times).

[0031] Step S2: Introducing SOC avg , dE and λ are three evaluation parameters, among which SOC avg Indicates each of the SOC end (i), dE represents the stability of each SOH (i), and λ represents the sum of the reliability scores of each SOH estimation result depending on the minimum SOC value. The values ​​of each evaluation parameter are calculated respectively by formulas (1)-(4):

[0032]

[0033] dE=2×[max(SOH(i))-mean(SOH(i))] (2)

[0034]

[0035]

[0036] Where: α(i) is the reliability score of the i-th SOH estimation result depending on the minimum SOC value.

[0037] In this step, dE is the preferred calculation formula of this embodiment, which is applicable to the case where the probabilities of max(SOH(i)) and min(SOH(i)) being abnormal are different, because min(SOH(i)) is more likely to have abnormal values ​​due to various factors. In actual applications, it is also possible that the probabilities of max(SOH(i)) and min(SOH(i)) being abnormal are the same. In this case, the calculation formula of dE can be adjusted to:

[0038] dE=max(SOH(i))-min(SOH(i))

[0039] In addition, dE can also be calculated using other calculation methods that characterize stability, such as the standard deviation formula std, the determination coefficient R-square after linear fitting, etc. The specific formula to be used is determined based on different application scenarios and the changing characteristics of SOH(i).

[0040] Step S3: SOC avgThe three evaluation parameters , dE, and λ are used to set the membership function respectively, thus forming the membership matrix A. The membership matrix A is a 3*3 matrix, and the three columns represent the three credibility types of SOH results: untrustworthy, generally credible, and very credible; the three rows represent SOC avg The membership degree of each credibility corresponding to the three evaluation parameters , dE, and λ. The matrix A is expressed as:

[0041]

[0042] Where: a1-a3 is SOC avg The membership of the three corresponding credibility classifications, b1-b3 are the membership of the three corresponding credibility classifications of dE, and c1-c3 are the membership of the three corresponding credibility classifications of λ. A clearer explanation is: a1 represents SOC avg is the probability of "untrustworthy", a2 represents SOC avg is the probability of "generally credible", a3 represents SOC avg The probability of "very credible" is given by [the sentence fragment "1"]. The membership function can be a linear trigonometric function, a rectangular function, a Gaussian function, etc., depending on the specific needs.

[0043] Step S4: SOC avg The weight coefficients of the three evaluation parameters , dE, and λ are set, and the probability density distribution matrix P of each credibility is calculated according to formula (5): j :

[0044]

[0045] Where: ω k For SOC avg The weight coefficients of the three parameters , dE, and λ satisfy k is the row number of matrix A; j is the column number of matrix A.

[0046] Matrix P j is a matrix with one row and three columns, which can be expressed as:

[0047] Pj = [d1d2d3]

[0048] Each column represents the probability density distribution result of each credibility level, that is, d1 represents the probability of "untrustworthy" in this calculation, d2 represents the probability of "generally trustworthy" in this calculation, and d3 represents the probability of "very trustworthy" in this calculation.

[0049] Step S5: Matrix P jThe reliability category of the position corresponding to the maximum probability density value is determined as the reliability classification calculation result of the current SOH estimation value SOH(n), thereby confirming the stability and accuracy of the current SOH estimation value SOH(n). The current SOH estimation value SOH(n) is the nth SOH estimation value. The focus of this invention is to introduce SOC avg , dE, and λ, and design probability density distribution value calculation methods with different credibility to assist in evaluating whether the data of each SOH(i) is reliable, and then finally determine whether the current SOH estimation value SOH(n) is accurate and reliable.

[0050] In order to simplify the calculation, the credibility K can be used to represent the matrix P j The credibility classification calculation result, that is, K=1 represents untrustworthy, K=2 represents generally trustworthy, and K=3 represents very trustworthy. By comparing the credibility K obtained in this evaluation with the credibility K0 obtained in the previous evaluation, the battery system SOH1 after this evaluation is obtained. Specifically, this embodiment sets that if K>K0, or K=K0 and the duration exceeds the set value, the battery system SOH1 after this evaluation is equal to the current SOH estimated value SOH(n); if K<K0, the battery system SOH1 after this evaluation is equal to the battery system SOH0 after the previous evaluation. In addition, when K<K0 and the duration exceeds the set value, or K=K0 and the duration exceeds the set value, a fitting formula can be used to fit SOH(i) and time D(i) within a certain time period, and the latest value obtained by fitting replaces the previous fitted old value for update. Fitting formulas include but are not limited to: linear, exponential, polynomial, mixed formulas, etc. It can be seen that the battery system SOH1 is an evaluation value with extremely high stability and reliability after calibration, which can accurately reflect the health status of the vehicle battery system.

[0051] The above is only a specific implementation of the present invention, but the design concept of the present invention is not limited to this. Any non-substantial changes to the present invention using this concept shall be deemed as an infringement of the protection scope of the present invention.

Claims

1. A method for enhancing the stability of battery health diagnosis results, characterized by: The steps include: Step S1: Obtain the number of SOH estimations of the battery system within a set time period n, record the SOH estimation results SOH(i) of each time, the corresponding date D(i) and the lowest SOC value SOC of the battery data used on that day end (i); Step S2: Introducing SOC avg , dE and λ are three evaluation parameters, among which SOC avg Indicates each of the SOC end (i), dE represents the stability of each SOH (i), and λ represents the sum of the reliability scores of each SOH estimation result depending on the minimum SOC value. The values ​​of each evaluation parameter are calculated respectively by formulas (1)-(4): dE=2×[max(SOH(i))-mean(SOH(i))] (2) Where: α(i) is the reliability score of the i-th SOH estimation result depending on the minimum SOC value; Step S3: SOC avg The three evaluation parameters , dE, and λ are used to set the membership function respectively, thus forming the membership matrix A. The membership matrix A is a 3*3 matrix, and the three columns represent the three types of credibility of the SOH results: untrustworthy, generally credible, and very credible; The 3 lines represent SOC avg , dE, and λ are the membership degrees of each credibility corresponding to the three evaluation parameters; Step S4: SOC avg The weight coefficients of the three evaluation parameters , dE, and λ are set, and the probability density distribution matrix P of each credibility is calculated according to formula (5): j : Where: ω k For SOC avg , dE, λ are the weight coefficients of the three parameters; k is the row number of matrix A; j is the column number of matrix A; Step S5: Matrix P j The credibility category of the position corresponding to the maximum probability density value is determined as the credibility classification calculation result of the current SOH estimation value SOH(n), thereby confirming the stability and accuracy of the current SOH estimation value SOH(n).

2. The method for enhancing the stability of battery health diagnosis results according to claim 1, characterized in that: In step S5, the credibility K is used to represent the matrix P j The credibility classification calculation result is that K=1 represents untrustworthy, K=2 represents generally trustworthy, and K=3 represents very trustworthy. By comparing the credibility K obtained in this evaluation with the credibility K0 obtained in the previous evaluation, the battery system SOH1 after this evaluation is obtained.

3. The method for enhancing the stability of battery health diagnosis results according to claim 2, characterized in that: If K>K0, or K=K0 and the duration exceeds the set value, the battery system SOH1 after this evaluation is equal to the current SOH estimation value SOH(n); if K<K0, the battery system SOH1 after this evaluation is equal to the battery system SOH0 after the previous evaluation.

4. The method for enhancing the stability of battery health diagnosis results according to claim 1, wherein: In step S1 , SOH(i) is the i-th SOH value of the available capacity of the battery system or the SOH value of the available capacity of a single cell in the battery system.

5. The method for enhancing the stability of battery health diagnosis results according to claim 1, wherein: In step S1 , the time period is set to a specified time length or a time period corresponding to when the estimated frequency reaches a specified number of times.

6. The method for enhancing the stability of battery health diagnosis results according to claim 1, wherein: In step S2 , the membership function may be a linear trigonometric function, a rectangular function, or a Gaussian function.

Images