Multi-metric fusion evaluation method for inconsistency of parallel battery packs

Abstract

The invention discloses a multi-metric fusion evaluation method for the inconsistency of parallel battery packs. The method comprises the steps of obtaining the average difference and the instantaneous maximum difference of dynamic performance parameters in the discharging process of multiple parallel battery packs; redundant features are removed by using a grey correlation analysis method, and evaluation indexes are simplified; setting an appropriate threshold value for each evaluation index, and calculating an inconsistency score shown by the battery pack through each evaluation index in the discharging process; an objective weight and a subjective weight are calculated by using a CRITIC method and an AHP method respectively, and then the subjective weight and the objective weight are fused through a geometric averaging method to obtain a comprehensive weight of each index in reflecting inconsistency; and finally, calculating the comprehensive inconsistency score of the battery pack according to the individual score of each evaluation index in combination with the corresponding weight, and dividing the comprehensive inconsistency score into four grades. According to the method, a multi-index and subjective and objective weight fused inconsistency quantitative evaluation system is constructed, and an accurate and reliable reference basis is provided for performance evaluation after the batteries are connected in parallel and grouped.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery packs, and in particular relates to a multi-metric fusion evaluation method for inconsistency of parallel battery packs. Background Art

[0002] In modern energy storage systems, parallel battery packs are widely used in electric vehicles, renewable energy storage equipment, and power backup systems due to their advantages in improving system capacity and stability. However, due to differences in manufacturing process, material properties, operating conditions, and aging among individual cells within a battery pack, these differences can lead to inconsistencies in the overall performance of the battery pack when used in parallel. This inconsistency not only reduces the overall efficiency and service life of the battery pack but can also pose safety risks, especially under high-load discharge conditions.

[0003] Currently, quantitative research on battery pack inconsistencies mainly focuses on series battery packs, while research on the inconsistency of parallel battery packs is relatively rare. Most existing studies conduct qualitative analysis of battery pack inconsistencies from a single perspective such as current, temperature or SOC, ignoring the comprehensive quantitative evaluation of multi-dimensional parameters. In addition, current evaluation indicators mostly rely on the differences in single-cell performance parameters such as capacity, impedance and voltage before the battery is grouped, and rarely involve the differences in intra-group distribution after parallel grouping. In contrast, the performance indicators of single-cell batteries before grouping may not fully reflect the inconsistencies in the actual use of parallel battery packs, resulting in insufficient reliability of the evaluation results and a high error rate. Therefore, how to achieve multi-dimensional and accurate quantitative evaluation of the inconsistency of parallel battery packs has become a key technical challenge for improving battery pack performance and reliability. Summary of the Invention

[0004] In order to overcome the above shortcomings, the present invention provides a multi-metric fusion evaluation method for the inconsistency of parallel battery packs.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is: A multi-metric fusion evaluation method for inconsistency of parallel battery packs is proposed. The specific steps are as follows: S1. Obtain the average difference and instantaneous maximum difference of each dynamic performance parameter within multiple parallel battery groups during constant current discharge; S2. Calculate the correlation coefficient between the average difference and the instantaneous maximum difference of each parameter in step S1 based on the grey correlation analysis method, eliminate redundant features, and obtain four evaluation indicators that can fully reflect the inconsistency of the parallel battery pack; S3. Calculate the objective weight using the CRITIC method, calculate the subjective weight using the AHP method, and fuse the subjective and objective weights using the geometric mean method to obtain the comprehensive weights of the four evaluation indicators in step S2 in reflecting inconsistency; S4. Setting appropriate thresholds for the four evaluation indicators in step S2 to calculate the inconsistency scores of the battery pack as shown by each evaluation indicator during discharge, which are recorded as separate scores for each evaluation indicator. S5. Finally, the comprehensive inconsistency score of the battery pack is obtained based on the individual scores of each evaluation indicator and the corresponding weights, and is divided into four levels.

[0006] Further optimization, the dynamic performance parameters selected in step S1 include temperature, current and SOC.

[0007] Further optimization, the step S1 specifically includes: S11, obtain the maximum and minimum value sets of each dynamic parameter in the group at each moment during the discharge process, respectively F max-i =max{F 1-i ,F 2-i ...F n-i},i=1,2,3...k F min-i =min{F 1-i ,F 2-i ...F n-i},i=1,2,3...k Where, F n-i is the parameter value of the nth battery in the group at time i, k is the total time, F represents the dynamic parameters of the selected group, F max-i 、F min-i are the maximum and minimum values of each parameter in the group at each moment during the discharge process; S12, using the root mean square error σ of the extreme value curve F and the maximum instantaneous difference within the group δ F To measure the average difference σ of each dynamic parameter in the parallel battery pack during the entire discharge process F and the instantaneous maximum difference δ F δ F =max{F max-i -F min-i}.

[0008] Further optimization, the step S2 specifically includes: S21, the average difference σ collected from multiple parallel battery groups F and the instantaneous maximum difference δ F It forms two sequences, X and Y, and normalizes the sequences to remove dimension. In the formula, x is the sequence to be normalized, x i is the data in the sequence, x i ′ is the normalized data; S22, calculate the grey relational degree between sequences X and Y, the value range is [0,1] Where, ξ is the grey relational degree between data sequences X and Y, n is the length of the sequence, x i ′ and y i ′ are the i-th and j-th data points after normalization of the two sequences, τ is the resolution coefficient, which is used to balance the influence of differences and is generally set to 0.5; S23. Eliminate redundant features based on the grey correlation degree ξ.

[0009] Further optimization, the fusion formula of subjective weight and objective weight in step S3 is:

[0010] Further optimization, the method for calculating the individual scores of the inconsistency of each evaluation index in step S4 is: Where, F j 、T hj Represent the characteristic values of each indicator and the corresponding threshold, s j Represents the individual score of each indicator, with a score range of 0 to 100, S = [s1,···,s j ,···,s n ].

[0011] Further optimization, the comprehensive inconsistency score G of the battery pack is obtained in step S5 Total The calculation method is G Total =W c S = ω c1 ·s1+···ω cj ·s j ···+ω cn ·s n .

[0012] Further optimization, the four grades in step S5 are excellent, good, medium and poor according to the scores from high to low.

[0013] The beneficial effects of the present invention are: First, starting from the multiple dynamic performance parameters of batteries after being connected in parallel, the average difference and instantaneous difference during the discharge process of the battery pack are comprehensively considered, and the grey correlation analysis method is used to eliminate redundant evaluation indicators. While ensuring a comprehensive evaluation of the inconsistency of the battery pack, the evaluation system is also streamlined. Secondly, the comprehensive weight of each evaluation indicator is combined with objective data and subjective judgment is also taken into account, avoiding the limitations of a single weighting method and making the weight distribution more reasonable and reliable. Compared with the inconsistency evaluation method of a single indicator and a single weight coefficient, the present invention has strong robustness and a low error rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flowchart of the multi-metric fusion evaluation method for quantifying inconsistency of parallel battery packs according to the present invention; Figure 2 The distribution curves of temperature, current and SOC of each branch battery during the discharge process of five parallel battery packs with different configurations; Figure 3 is the ohmic internal resistance R0 of each single cell in battery group 1, group 2 and group 3. DETAILED DESCRIPTION

[0015] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is described in detail below in conjunction with specific embodiments. The following embodiments are implemented based on the technical solutions of the present invention, and provide detailed implementation methods and specific operating procedures. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the following embodiments.

[0016] A multi-metric fusion evaluation method for inconsistency of parallel battery packs is proposed. The specific steps are as follows: S1. Under the discharge conditions of 25°C and 1C, obtain the average difference σ in temperature, current and SOC within the group during the constant current discharge of 12 parallel battery packs. F and the instantaneous maximum difference δ F ,The extracted eigenvalues are shown in Table 1; F max-i =max{F 1-i ,F 2-i ...F n-i},i=1,2,3...k F min-i =min{F 1-i ,F 2-i ...F n-i},i=1,2,3...k δ F =max{F max-i -Fmin-i} Where, F n-i is the parameter value of the nth battery in the group at time i, k is the total time, F represents the dynamic parameters of the selected group, F max-i 、F min-i are the maximum and minimum values of each parameter in the group at each moment during the discharge process.

[0017] Table 1 Characteristic values of dynamic performance parameters extracted from 12 battery packs Battery pack number <![CDATA[δ T (℃)]]> <![CDATA[σ T (℃)]]> <![CDATA[δ I (A)]]> <![CDATA[σ I (A)]]> <![CDATA[δ SOC (%)]]> <![CDATA[σ SOC (%)]]> Pack_#1 4.8 1.05 4.71 1.26 9.89 7.24 Pack_#2 0.7 0.34 1.24 1.01 3.11 2.20 Pack_#3 0.7 0.42 1.07 0.27 1.87 1.44 Pack_#4 2.1 0.85 2.02 0.68 4.06 3.02 Pack_#5 3.5 1.36 3.52 1.26 6.97 5.08 Pack_#6 1.5 1.24 3.73 0.71 6.62 5.06 Pack_#7 2.0 1.02 1.60 0.42 5.08 3.6 Pack_#8 1.0 0.74 1.19 0.22 1.86 1.33 Pack_#9 1.4 0.73 1.69 0.39 3.53 2.52 Pack_#10 1.6 0.75 3.28 0.96 6.24 4.57 Pack_#11 3.2 1.43 3.44 1.32 6.46 4.81 Pack_#12 1.6 1.12 2.52 0.56 6.36 4.63

[0018] S2. Calculate the correlation coefficients between the average differences and the instantaneous maximum differences of the temperature, current, and SOC within the group described in step S1 using the grey correlation analysis method, remove redundant features, and simplify the evaluation indicators, specifically: S21, the average difference σ collected from multiple parallel battery groups F and the instantaneous maximum difference δ F It forms two sequences, X and Y, and normalizes the sequences to remove dimension. In the formula, x is the sequence to be normalized, x i is the data in the sequence, x i ′ is the normalized data; S22, calculate the grey relational degree between sequences X and Y, with the value range being [0,1]. The calculation result is as follows As shown in Table 2; Where, ξ is the grey relational degree between data sequences X and Y, n is the length of the sequence, x i ′ and y i ′ are the i-th and j-th data points after normalization of the two sequences, τ is the resolution coefficient, which is used to balance the influence of differences and is generally set to 0.5; S23. Eliminate redundant features based on the grey correlation degree ξ.

[0019] Table 2 Grey relational degree between the average difference and instantaneous maximum difference of each parameter Dynamic parameters temperature Current SOC Grey relational degree ξ 0.59 0.72 0.70

[0020] Considering ξ≥0.7, the two indicators are strongly correlated in the mapping degree of battery pack inconsistency. Since the instantaneous maximum difference can better reflect the inconsistency within the group than the average difference, the instantaneous maximum difference between current and SOC is retained due to the strong correlation between the average difference and the instantaneous maximum difference between current and SOC. I and δ SOCUsed to reflect the degree of inconsistency of the battery pack in the corresponding parameters.

[0021] So we use the average difference of temperature within the group σ T and the instantaneous maximum difference δ T , the instantaneous maximum difference δ between current and SOC I and δ SOC These four evaluation indicators are used to reflect the inconsistency of parallel battery packs.

[0022] S3. Calculate the objective weight using the CRITIC method, calculate the subjective weight using the AHP method, and combine the subjective and objective weights using the geometric mean method to obtain the comprehensive weight of each indicator in reflecting inconsistency, as follows: S31. Use the CRITIC method to calculate the objective weight of each indicator S311: Normalize the eigenvalue data of the 12 groups of four evaluation indicators in Table 1 to obtain the normalized matrix X′ where x′ mn It represents the nth evaluation index value of the mth battery pack after normalization; S312. Calculate the standard deviation of each evaluation indicator: For each indicator j, calculate its standard deviation The standard deviation reflects the degree of variation of indicator j among different battery packs: where x′ ij It represents the jth evaluation index value of the i-th battery pack after normalization. It represents the average value of the j-th evaluation index in all m battery packs, that is Reflects the overall average level of the indicator; S313. Calculate the correlation of each evaluation indicator: The correlation between the standards is used to measure the similarity between the two standards. It is usually calculated by the Pearson Correlation Coefficient. For the correlation between the indicators j and s, r js , the calculation formula is as follows: Where x′ is and represents the corresponding value of another evaluation index (s≠j), and m is the number of battery packs; S314. Calculate objective weight: The larger the standard deviation of the evaluation index, the greater the amount of information and the greater the weight; the higher the correlation, the higher the information redundancy and the smaller the weight. The weight of each evaluation index is calculated by combining the standard deviation and the correlation between the indicators. The calculation formula is as follows: in, is the standard deviation of the j-th evaluation index, r js is the correlation between indicators j and s; Finally, normalize the weights of all criteria to ensure that the sum of all weights is 1 The results of calculating the objective weights of various indicators using the CRITIC method are shown in Table 5.

[0023] S32. Use the AHP method to calculate the subjective weight of each indicator S321. Set the hierarchy: The hierarchical structure is divided into a target layer, a criterion layer, and a solution layer, wherein the solution layer is the parallel battery pack to be evaluated, the criterion layer is the evaluation index of the inconsistency of the parallel battery pack described in step S2, and the target layer is the comprehensive evaluation score of the battery pack inconsistency; S322. Establish a judgment matrix: When establishing a judgment matrix, decision makers need to compare the relative importance of each indicator and score it according to the 9-level scale shown in Table 3. i ,a j (i,j=1,2,3,…,n) are the indicators in the criterion layer, then a ij Representation factor a i and factor a j In contrast, factor a i The importance of the impact on the performance of the battery pack itself, all a ij Composed of judgment matrix P, where a ij =1 / a ji Table 3 Importance levels and scale values The relative importance of the two factors on battery pack performance Scale value <![CDATA[Factor a i is equally important as Factor a j > 1 <![CDATA[Factor a i is more important than Factor a j slightly]]> 3 <![CDATA[Factor a i is more important than Factor a j significantly]]> 5 <![CDATA[Factor a i more important than Factor a j strongly important]]> 7 <![CDATA[Factor a i is more important than Factor a j is extremely important]]> 9 Intermediate value, used to fine-tune the judgment intensity 2,4,6,8 Taking the severity of the impact on the battery pack's own performance as a reference, the four indicators in the criterion layer are compared two by two, and the judgment matrix P is constructed in combination with Table 3. S323. Calculate subjective weight: According to the judgment matrix P, the maximum characteristic root λ can be calculated max And the corresponding normalized eigenvector ω′, by calculating, λ max=4,ω′=(0.4445,0.1111,0.2963,0.1481); S324, consistency test: From Table 4, we can see that when n=4, λ max =4, RI = 1.24, CR = 0 < 0.1 can be calculated by the following formula, indicating that the judgment matrix P has good consistency

[0024] Table 4 Standard value RI of average random consistency signal n 1 2 3 4 5 6 7 RI 0 0 0.58 0.90 1.12 1.24 1.32

[0025] Therefore, the normalized components of ω′ can be used as subjective weight coefficients of inconsistency evaluation indicators, as shown in Table 5.

[0026] S33. Calculate the comprehensive weight of the integration of subjective and objective factors using the geometric mean method: The objective weight vector obtained by the CRITIC method is recorded as W o =[ω o1 ,ω o2 ,···,ω on ]; The subjective weight vector obtained by AHP method is recorded as W s =[ω s1 ,ω s2 ,···,ω sn ]; the comprehensive weight vector of the fusion of subjective and objective factors is recorded as W c =[ω c1 ,ω c2 ,···,ω cn ].

[0027] The objective and subjective weights of each evaluation index are fused by the geometric mean method to obtain the weight proportion of each index in describing the inconsistency of the battery pack. The comprehensive weights of the four evaluation indicators are calculated by the following fusion formula as shown in Table 5:

[0028] Table 5 Weight coefficients of the four inconsistency evaluation indicators Evaluation indicators <![CDATA[δ T ]]> <![CDATA[σ T ]]> <![CDATA[δ I ]]> <![CDATA[δ SOC ]]> Objective weight (%) 25.23 34.99 21.68 18.10 Subjective weight (%) 44.45 11.11 29.63 14.81 Comprehensive weight (%) 35.28 20.77 26.70 17.25

[0029] S4. Calculate the individual scores for the inconsistency of each evaluation indicator S41. Set thresholds for the four evaluation indicators: Through research and analysis, appropriate thresholds are set for the four evaluation indicators proposed in this invention, as shown in Table 6.

[0030] Table 6 Threshold settings for the four evaluation indicators Evaluation indicators <![CDATA[δ T ]]> <![CDATA[σ T ]]> <![CDATA[δ I ]]> <![CDATA[δ SOC ]]> Threshold 5℃ 2℃ <![CDATA[150%I ave ]]> 10％ Among them, I ave is the average current of each battery in the parallel battery pack Where, I total is the total current flowing through the battery pack, and n is the number of batteries connected in parallel in the pack.

[0031] S42. Calculate the individual scores for the inconsistency of the four evaluation indicators, denoted as S = [s1, s2, s3, s4].

[0032] S5. Calculate the comprehensive inconsistency score of the parallel battery pack and classify its inconsistency level The weight vector W obtained in step S3 c Multiply the individual scoring vectors S of the inconsistency of each evaluation index obtained in step S4 to calculate the comprehensive score G of the inconsistency of the parallel battery pack. Total : G Total =35.28%·s1+20.77%·s2+26.70%·s3+17.25%·s4.

[0033] To verify the accuracy of the evaluation method proposed in this invention, five three-parallel battery packs with different battery configurations were set up for experimental verification, as shown in Table 7. The individual cells were grouped by capacity. Group 1 consisted of three new cells (100% SOH) with similar cell capacities connected in parallel; Groups 2 and 3 also consisted of three cells with similar cell capacities connected in parallel, but with cell SOHs of approximately 88% and 70%, respectively; Group 4 consisted of three cells with a 7% difference in SOH; and Group 5 consisted of three cells with a 13% difference in SOH.

[0034] Table 7 Five groups of three-in-one battery packs with different configurations

[0035] The multi-metric fusion evaluation method for inconsistency proposed in this invention is used to detect the five parallel battery packs with different configurations in Table 7. Figure 2 The temperature, current, and SOC distribution curves of each branch battery during the discharge process of five parallel battery packs with different configurations are given, and the evaluation results are shown in Table 8. It can be seen that although Group 1, Group 2, and Group 3 are all composed of three batteries of the same capacity connected in parallel, the evaluation system's assessment of their inconsistency is 22.78 points (excellent), 36.32 points (good), and 52.25 points (medium), respectively. In order to further verify the evaluation results, the ohmic internal resistance R0 of each single cell in Battery Packs 1, 2, and 3 was measured, as shown in Figure 8. Figure 3As shown in the figure, it can be seen that although the single cell capacities of the three battery packs are similar, the internal resistance of the single cells in the packs are different.

[0036] Table 8 Comprehensive quantitative evaluation results of inconsistency of five parallel battery packs with different configurations

[0037] Furthermore, as shown in Table 9, the coefficient of variation of the parameter distribution within each group was calculated using the following formula: This is used to quantify the distribution differences in the capacity and internal resistance of individual batteries within the group, and then cross-check and verify the coefficient of variation and scoring results of each battery group. Where: is the coefficient of variation, which describes the distribution difference of the parameters within the group; σ and u are the standard deviation and mean of the corresponding parameters, respectively.

[0038] Table 9 Differences in parameter distribution within the group and cross-validation of evaluation results battery pack Group 1 Group 2 Group 3 Group 4 Group 5 Internal resistance variation coefficient 0.02 0.08 0.13 0.12 0.11 Capacity variation coefficient 0 0.01 0.02 0.07 0.14 Inconsistency score 22.78 36.32 52.25 57.67 79.29 Inconsistency level excellent good medium medium Difference

[0039] Table 9 shows that the coefficient of variation of the internal resistance of a battery in Group 1 was 0.02, 0.08 for Group 2, and 0.13 for Group 3. This is consistent with the evaluation score; that is, the greater the difference in internal resistance distribution, the higher the inconsistency score, preliminarily indicating that the evaluation system can accurately reflect the actual battery pack situation. Inconsistency is also correlated with capacity distribution differences. Group 5 had the highest capacity coefficient of variation of 0.14 among the five groups, resulting in an inconsistency score of 79.29, rated poor. Group 4 had a capacity distribution difference of 0.07, resulting in an inconsistency score of 57.67, rated moderate. Group 1 had a capacity distribution difference of 0, resulting in the lowest inconsistency score, rated excellent. This demonstrates that the evaluation system can comprehensively reflect inconsistencies caused by differences in multiple battery pack parameters.

[0040] Furthermore, because the capacity and resistance of individual cells do not decay synchronously, even if the battery packs have the same capacity, battery packs in poor health may exhibit greater inconsistency when discharged in parallel. Therefore, it is reasonable to classify the inconsistency levels of Groups 1, 2, and 3 into different levels. For Groups 4 and 5, the maximum differences in the SOH of their individual cells were 14% and 25%, respectively, with inconsistency scores of 57.67 (medium) and 79.29 (poor), respectively. These scores were positively correlated with the differences between individual cells within the group, further validating the rationality of the evaluation system.

[0041] In summary, the five typical battery packs in Table 7 contain configuration differences in various situations when the batteries are grouped. The method proposed in the present invention can accurately identify their inconsistencies, which confirms the rationality of the proposed method.

[0042] The above shows and describes the main features, methods of use, basic principles, and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and improvements may be made to the present invention based on actual circumstances without departing from the spirit and scope of the present invention. Such changes and improvements are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-metric fusion evaluation method for inconsistency of parallel battery packs, characterized by: The specific steps are as follows: S1. Obtain the average difference and instantaneous maximum difference of each dynamic performance parameter within multiple parallel battery groups during constant current discharge; S2. Calculate the correlation coefficient between the average difference and the instantaneous maximum difference of each parameter in step S1 based on the grey correlation analysis method, eliminate redundant features, and obtain four evaluation indicators that can fully reflect the inconsistency of the parallel battery pack; S3. Calculate the objective weight using the CRITIC method, calculate the subjective weight using the AHP method, and fuse the subjective and objective weights using the geometric mean method to obtain the comprehensive weights of the four evaluation indicators in step S2 in reflecting inconsistency; S4. Setting appropriate thresholds for the four evaluation indicators in step S2 to calculate the inconsistency scores of the battery pack as shown by each evaluation indicator during discharge, which are recorded as separate scores for each evaluation indicator. S5. Finally, the comprehensive inconsistency score of the battery pack is obtained based on the individual scores of each evaluation indicator and the corresponding weights, and is divided into four levels.

2. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: The dynamic performance parameters selected in step S1 include temperature, current and SOC.

3. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: The step S1 specifically includes: S11, obtain the maximum and minimum value sets of each dynamic parameter in the group at each moment during the discharge process, respectively F max-i =max{F 1-i ,F 2-i ...F n-i },i=1,2,3...k F min-i =min{F 1-i ,F 2-i ...F n-i },i=1,2,3...k Where, F n-i is the parameter value of the nth battery in the group at time i, k is the total time, F represents the dynamic parameters of the selected group, F max-i 、F min-i are the maximum and minimum values of each parameter in the group at each moment during the discharge process; S12, using the root mean square error σ of the extreme value curve F and the maximum instantaneous difference within the group δ F To measure the average difference σ of each dynamic parameter in the parallel battery pack during the entire discharge process F and the instantaneous maximum difference δ F δ F =max{F max-i -F min-i }。 4. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: The step S2 specifically includes: S21, the average difference σ collected from multiple parallel battery groups F and the instantaneous maximum difference δ F It forms two sequences, X and Y, and normalizes the sequences to remove dimension. In the formula, x is the sequence to be normalized, x i is the data in the sequence, x′ i is the normalized data; S22, calculate the grey relational degree between sequences X and Y, the value range is [0,1] Where ξ is the grey relational degree between data sequences X and Y, n is the length of the sequence, and x′ i and y′ i are the i-th and j-th data points after normalization of the two sequences, respectively. τ is the resolution coefficient, which is used to balance the influence of differences and is generally set to 0.

5. S23. Eliminate redundant features based on the grey correlation degree ξ.

5. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: The fusion formula of subjective weight and objective weight in step S3 is:

6. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: The method for calculating the individual scores of the inconsistencies of the evaluation indicators in step S4 is: Where, F j 、T hj Represent the characteristic values of each indicator and the corresponding threshold, s j Represents the individual score of each indicator, with a score range of 0 to 100, S = [s1,···,s j ,···,s n ].

7. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: In step S5, the comprehensive inconsistency score G of the battery pack is obtained. Total The calculation method is G Total =W c ·S=ω c1 ·s1+···ω cj ·s j ···+ω cn ·s n Among them, W c is the vector set of subjective and objective fusion weights of each evaluation index in evaluating battery pack inconsistency, ω c1 、ω cj 、ω cn is the fusion weight of the 1st, jth, and nth evaluation indicators, S is the vector set of the individual scores of the inconsistency of the battery pack under test evaluated by each evaluation indicator, s1, s j 、s n A separate score for the battery pack to be tested based on the evaluation indicators 1, j, and n to assess the inconsistency of the battery pack.

8. The multi-metric fusion evaluation method for inconsistency of a parallel battery pack according to claim 1, characterized in that: The four grades in step S5 are excellent, good, medium and poor according to the scores from high to low.

Images