CN113484764B - Retired battery SOH and consistency assessment method based on multidimensional impedance spectrum - Google Patents

Abstract

The invention discloses an evaluation method of SOH and consistency of retired batteries based on multi-dimensional impedance spectroscopy, which utilizes a fingerprint identification algorithm to identify multi-dimensional EIS curve clusters of batteries to be evaluated, and finds out multi-dimensional EIS curve clusters matched with the multi-dimensional EIS curve clusters in a fingerprint spectrum library formed by multi-dimensional EIS curve clusters of sample batteries, so that SOH and SOC states of the batteries to be evaluated can be determined; dividing the batteries with SOH consistency into a group; and then the similarity between the maps is used for further evaluation of consistency in the batteries in the group. The invention utilizes the multidimensional EIS curve to represent the retired battery state, has higher accuracy, and considers that the performance can still have larger deviation under the condition of consistent SOH of the batteries, so that the consistency evaluation is carried out by utilizing two indexes of the SOH and the similarity of the battery fingerprints in the same group, and the evaluation result has higher reliability.

Description

Evaluation Method of SOH and Consistency of Decommissioned Batteries Based on Multidimensional Impedance Spectroscopy

technical field

The invention belongs to the technical field of batteries, in particular to a method for evaluating the SOH and consistency of decommissioned batteries based on multidimensional impedance spectroscopy.

Background technique

In recent years, electric vehicles have developed rapidly due to their energy saving, emission reduction and environmental friendliness, and the rapid development of electric vehicles has led to a sharp increase in the output of power batteries. However, when the capacity of the vehicle's power battery decays to 80%, it will not be able to meet the power requirements of electric vehicles and will face retirement, which will lead to the consumption of decommissioned power batteries. From the perspective of the battery, if the decommissioned power battery is directly scrapped and disassembled, the service life of the battery will be seriously shortened, the energy utilization efficiency will be reduced, and resources will be seriously wasted. Cascade utilization can solve this problem very well. Through cascade utilization, the performance of the power battery can be fully utilized, and the cost can also be reduced.

The factors involved in battery life attenuation are relatively complex, and the consistency of retired batteries varies greatly. In order to ensure the safety of cascaded battery systems and improve the performance and service life of battery packs, decommissioned power batteries must undergo SOH and consistency evaluations before they can be reassembled Utilization; Accurate battery SOH and consistency evaluation methods can improve the production efficiency of battery cascade utilization enterprises and reduce the output rate of unqualified products. Therefore, it is particularly important to accurately evaluate the SOH and consistency of decommissioned power batteries.

Most of the SOH and consistency evaluation methods in the related art evaluate the consistency based on the open circuit voltage, capacity, and internal resistance of the battery. This method only judges based on the external characteristics of the battery, and has the characteristics of simple operation but low accuracy. There are also factors such as battery charge and discharge curves for evaluation, but the test time is long, and it is impossible to quickly and accurately complete the consistent identification of large quantities of batteries, which is not conducive to commercial use. In addition, there is an evaluation method based on electrochemical AC impedance spectroscopy in related technologies, but its EIS curve is measured at room temperature and fully charged, which can only reflect the impedance state of the battery under a certain condition, and cannot accurately reflect the The real condition of the battery, the reliability of the consistency evaluation results is low.

Contents of the invention

The present invention aims to solve the shortcomings of the above-mentioned prior art, and proposes a method for evaluating the SOH and consistency of decommissioned batteries, in order to use the two indicators of SOH and fingerprint similarity to achieve consistency evaluation, thereby improving the evaluation In order to solve the problems of long evaluation time and low precision existing in existing methods.

The present invention adopts following technical scheme in order to achieve the above-mentioned purpose of the invention:

A method for evaluating the SOH and consistency of decommissioned batteries based on multidimensional impedance spectroscopy in the present invention is characterized in that it includes:

S1. Randomly select several batteries, and measure the SOH of each battery by using a typical cycle charging and discharging method under a constant temperature state, and then measure the multidimensional EIS curve cluster of each battery under different SOC states; repeat the test after changing the temperature process, so as to obtain the multi-dimensional EIS curve cluster library of all batteries at different SOC states and temperatures T to form a sample battery spectrum library;

S2. Obtain a cluster of EIS curves under excitation currents of different amplitudes with the real and imaginary parts of the impedance as coordinate axes of the battery to be evaluated and use it as its spectrum, and match it with the spectrum in the sample battery spectrum library;

If the matching is successful, the SOH and SOC states corresponding to the successfully matched spectra are used as the SOH and SOC states of the battery to be evaluated;

If the matching fails, measure the SOH of the battery to be evaluated and the multidimensional EIS curve clusters under different SOC and temperature T conditions according to the process of step S1, and add them to the sample battery library;

S3. According to the process of step 2, match all the batteries to be evaluated;

S4. Taking SOH as the first index of consistency evaluation, setting the value range of SOH, and grouping all the batteries to be evaluated within the value range into one group;

S5. Number the batteries to be evaluated in the same group, and use No. 1 battery to be evaluated as a reference battery, and calculate the similarity between the fingerprints of other batteries to be evaluated and the fingerprints of the reference battery;

S6. Taking the similarity as the second index of the consistency evaluation, judge the battery to be evaluated whose similarity is higher than the set threshold as the battery with consistency, and classify it into one category, and the remaining batteries to be evaluated follow step S5 And the process of step S6 continues to sort until all the batteries to be evaluated are sorted.

The method for assessing the SOH and consistency of decommissioned batteries based on multidimensional impedance spectroscopy in the present invention is also characterized in that the spectrum matching method in the step S2 is a fingerprint recognition algorithm based on minutiae patterns or fingerprint recognition algorithms based on texture patterns.

The minutiae-based pattern fingerprint recognition algorithm is to convert each spectrum in the sample battery spectrum database and the battery spectrum to be evaluated into respective point sets composed of feature points, and calculate each feature in the point set transformed from the battery spectrum to be evaluated The distance between the points and the corresponding feature points in the point set converted from each map in the map library. If the distance is less than the set point offset distance threshold, it means that the corresponding two feature points match, and the point set of battery conversion to be evaluated is counted. The number of feature points matched with the converted point sets of each map in the map library, and the map with the largest number of matching points is used as a candidate map. If the number of matching points is greater than the set threshold, the candidate map is successfully matched spectrum; otherwise, it means that the matching failed.

The fingerprint recognition algorithm based on the texture mode is to convert each spectrum in the sample battery spectrum library and the spectrum of the battery to be evaluated into a line set composed of lines, and calculate the distance between adjacent lines in each line set. The dynamic bending distance is used as the feature vector of the map, and then the distance between the feature vector of the battery to be evaluated and the feature vectors of each map in the sample battery map library is calculated, and the minimum distance is selected to determine whether it is less than the set value. If it is less than the specified map offset distance threshold, the map corresponding to the minimum distance will be regarded as the map that matches successfully; otherwise, it means that the match fails.

Compared with prior art, the beneficial effect of the present invention is:

1. The present invention uses multidimensional EIS impedance spectroscopy to characterize the state of the battery, which can more comprehensively and accurately reflect the real situation of the battery compared to conventional characterization methods.

2. The present invention recognizes the fingerprints formed by the impedance spectrum, and converts the consistency evaluation into the similarity problem between the fingerprints, so that the consistency evaluation is more reasonable and intuitive.

3. The present invention is still applicable to the health status and consistency assessment of a large number of decommissioned power batteries, with short test period and high work efficiency.

4. The present invention adopts two indicators for consistency evaluation, which has higher reliability, which is beneficial to improve the overall performance of the recombined and utilized battery pack and prolong its service life.

Description of drawings

Fig. 1 is the schematic diagram of EIS curve of the present invention changing with SOC;

Fig. 2 is a schematic diagram of the EIS curve of the present invention as the number of charge-discharge cycles varies;

Fig. 3 is the fingerprint matching algorithm flowchart based on minutiae pattern of the present invention;

Fig. 4a is a three-dimensional EIS curve diagram that the present invention is used to specifically illustrate fingerprint recognition;

Fig. 4b is another three-dimensional EIS graph that the present invention is used for specifying fingerprint identification;

Fig. 5 is a flow chart of the method for determining the SOH and SOC of the battery to be evaluated according to the present invention.

Detailed ways

In this embodiment, the method for evaluating the SOH and consistency of decommissioned batteries based on multidimensional impedance spectroscopy includes:

S1. Randomly select several batteries, and measure the SOH of each battery by using a typical cycle charging and discharging method under a constant temperature state, and then measure the multidimensional EIS curve cluster of each battery under different SOC states; repeat the test after changing the temperature process, so as to obtain the multi-dimensional EIS curve cluster library of all batteries at different SOC states and temperatures T to form a sample battery spectrum library;

Specifically, the method for obtaining multidimensional EIS curve clusters is:

S1.1. Apply a sinusoidal current signal with frequency f ₁ , f ₂ , f ₃ ...f _b and amplitude i ₁ to the decommissioned battery;

S1.2. Synchronously sample the voltage, current, and temperature signals under the excitation signal, and after Fourier transform analysis, the impedance Z ₁ (f ₁ ), Z ₁ (f ₂ ), Z of the battery at each frequency can be obtained ₁ (f ₃ )… Z ₁ (f _b );

S1.3. From the real part and the imaginary part of the impedance, the EIS curve under the excitation signal whose amplitude is _i1 can be fitted at the current temperature;

S1.4. Change the amplitude of the excitation current to i ₂ , i ₃ ...i _a to get the impedance Z ₂ (f ₁ ), Z ₂ (f ₂ ), Z ₂ (f ₃ )...Z ₂ of the battery (f _b ), Z ₃ (f ₁ ), Z ₃ (f ₂ ), Z ₃ (f ₃ )...Z ₃ (f _b ),..., Z _a (f ₁ ), Z _a (f ₂ ), Z _a (f ₃ )…Z _a (f _b ) fit the corresponding EIS curves to form a cluster of multi-dimensional EIS curves under excitation currents of different amplitudes with the real part and imaginary part of the impedance as the coordinate axes of the decommissioned battery.

As shown in Figure 1 and Figure 2, it can be seen that with the change of SOC and cycle charge and discharge times, the EIS curve changes significantly. Here, the change of cycle charge and discharge times is used to replace the change of battery SOH.

S2. Obtain a cluster of EIS curves under excitation currents of different amplitudes with the real and imaginary parts of the impedance as coordinate axes of the battery to be evaluated and use it as its spectrum, and match it with the spectrum in the sample battery spectrum library;

Wherein, the map matching method is a minutiae pattern-based fingerprint recognition algorithm or a texture pattern-based fingerprint recognition algorithm.

Fingerprint identification refers to comparing the fingerprint feature of the sample to be identified with the fingerprint feature in the fingerprint library to find the corresponding or closest fingerprint feature. The simple mathematical description of the specific identification process is as follows: suppose there is m in the fingerprint library D state patterns, that is, S ₁ , S ₂ ,...S _m , and each state pattern S _i has an n-dimensional fingerprint feature, and the fingerprint feature of the sample to be identified is y=(y ₁ ,y ₂ ,...y _n ) ^T , then the fingerprint feature recognition is to find the _xi closest to y in the fingerprint database D, and the state mode corresponding to _xi is the state mode of y.

The fingerprint recognition algorithm based on the minutiae pattern is to convert each spectrum in the sample battery spectrum library and the battery spectrum to be evaluated into a respective point set composed of feature points, and calculate each feature point and the spectrum library in the point set transformed from the battery spectrum to be evaluated The distance of the corresponding feature points in the converted point set of each map in each map. If the distance is less than the set point offset distance threshold, it means that the corresponding two feature points match. The point set of the battery map to be evaluated and the map library are counted. The number of matching feature points between the point sets converted from each map in the map, and the map with the largest number of matching points is used as the candidate map. If the number of matching points is greater than the set threshold, the candidate map is the successful map. ; otherwise, the match failed.

Specifically, as shown in Figure 3, the battery impedances Z ₁ (f ₁ ), Z ₁ (f ₂ ), Z ₁ (f ₃ )...Z ₁ (f _b ) obtained during the process of obtaining the multidimensional EIS curve cluster, …Z _a (f ₁ ), Z _a (f ₂ ), Z _a (f ₃ )…Z _a (f _b ), as the feature points of the fingerprint, constitute the feature point set of the fingerprint, by calculating the feature points of the fingerprint The number of matching points between sets is used to judge whether the fingerprints match. The distance between feature points is obtained by the following formula (1):

d(Z _i (f _j ),Z _i ′(f _j ))=(Re _i (f _j )-Re′ _i (f _j )) ² +(Im _i (f _j )-Im′ _i (f _j )) ² (1)

d(Z _i (f _j ), Z′ _i (f _j )) indicates the distance between Z _i (f _j ) and Z′ _i (f _j ), and Z _i (f _j ) indicates the battery corresponding to point set A The impedance under the excitation current with amplitude i _i and frequency f _j , Z′ _i (f _j ) represents the impedance of the battery corresponding to point set B under the excitation current with amplitude i _i and frequency f _j ; Re _i (f _j ), Im _i (f _j ) are the real and imaginary parts of the impedance corresponding to Z _i (f _j ), respectively, Re′ _i (f _j ), Im′ _i (f _j ) are Z′ _i (f _j ) corresponds to the real and imaginary parts of the impedance.

Compare the distance between the calculated feature points with the distance threshold between points. If it is less than the threshold, it is considered that the two feature points match. If the number of matching points between the two point sets exceeds the threshold of the number of fingerprint matches, it is considered two The fingerprints match. The method of determining the distance threshold between points is as follows: when the battery is in the same SOH, SOC, and temperature T state, the multidimensional EIS curve cluster of the battery is obtained multiple times, and the distance between the corresponding feature points is calculated, and the average value is used as the value in this state. Threshold value; the judgment threshold values corresponding to other SOH, SOC and temperature T states are obtained in the same way. The quantity threshold is determined by the precision required by the system.

More specifically, Fig. 4a and Fig. 4b are used to illustrate the identification process. As shown in the figure, there are two to-be-evaluated impedance real and imaginary parts as the coordinate axes at excitation signal amplitudes I ₁ , I ₂ , and I ₃ The three-dimensional EIS curve, the points on each EIS curve are the impedance points corresponding to each frequency calculated during the fitting process. By calculating the distance between the corresponding feature points, such as the distance between the impedance points on the EIS curve whose excitation current amplitude is the same as I ₁ and the frequency is the same as f ₁ , it is judged whether the feature points match, and then the matching points are counted. The number is compared with the number threshold to determine whether the maps match.

The fingerprint matching algorithm based on the texture mode is to convert each map in the sample battery map library and the map of the battery to be evaluated into a line set composed of ridges, and calculate the dynamic bending between adjacent ridges in each line set The distance is used as the feature vector of the graph, and the distance between the feature vector of the graph of the battery to be evaluated and the feature vectors of each graph in the sample battery graph library is calculated, and the minimum distance is selected to determine whether it is less than the set graph deviation. If it is less than the shift distance threshold, the map corresponding to the minimum distance will be regarded as the matching successful map; otherwise, the matching will fail.

Specifically, the dynamic time wrapping (DTW) distance can effectively handle the case where there are local displacements in the sequence. The DTW distance constructs an alignment matrix and uses a dynamic programming method to find a curved path in two time series that minimizes the cumulative distance between the two time series. Define two time series as P=[p ₁ ,p ₂ ,...p _m ] ^T and Q=[q ₁ ,q ₂ ,...q _m ] ^T , and the sequence lengths are m and n respectively. In order to align P and Q using the DTW distance, first construct a distance matrix A with m rows and n columns, namely

In formula (2): element a _ij =d(p _i , q _j )=(p _i -q _i ) ² in A represents the alignment distance between time series points p _i and q _j . A curved path is a continuous set of feature maps of P and Q in A, denoted as W=[w ₁ ,w ₂ ,w _k ,...,w _K ]. The kth element in W is defined as w _k =(a _ij ) _k .

W needs to satisfy the following constraints:

Boundedness: max(m,n)≤K≤m+n-1.

Boundary: w ₁ =a ₁₁ and w _K =a _mn are used to represent the starting point and the ending point of W, respectively.

Continuity: For w _k ＝a _ij , its adjacent elements w _k-1 ＝a _i'j' satisfy i-i'≤1, j-j'≤1, this constraint limits the adjacent elements in W to be Adjacent unit in A.

Monotonicity: i-i'≥0, j-j'≥0.

There are multiple Ws in A that satisfy the above constraints. The DTW distance between P and Q refers to the W with the smallest cumulative distance. The objective function is expressed as formula (3):

In formula (3): f _DTW (P, Q) represents the DTW distance between P and Q; K represents the length of the minimum curved path; w _i is the i-th element in the minimum curved path. Use the dynamic programming algorithm to solve the DTW distance, and the recursive algorithm is expressed as formula (4):

In formula (4): D(i, j) represents the sum of the minimum cumulative value of element a _ij and the length of the curved path part of the preceding section.

After calculating the dynamic bending distance between the EIS curves, the eigenvectors of the fingerprints are composed, and the Euclidean distance between the eigenvectors is calculated again, and the calculation result is compared with the threshold value of the spectrum shift to judge whether the fingerprints match. The method of determining the shift threshold of the map is: when the battery is in the same SOH, SOC, and temperature T state, the multidimensional EIS curve of the battery is obtained multiple times, the distance between the corresponding feature vectors is calculated, and the average value is used as the threshold in this state , the judgment thresholds corresponding to other SOH, SOC and temperature T states are obtained in the same way.

More specifically, Fig. 4a and Fig. 4b are still used to illustrate the identification process, and the dynamic bending distances dtw ₁ and dtw 2 between the EIS curves corresponding to the amplitude excitations I ₁ and I ₂ , I ₂ and _{I 3 in} Fig. 4a are calculated, and the composition The eigenvector l _a =[dtw ₁ ,dtw ₂ ], similarly, the eigenvector l _b =[dtw ₁ ′,dtw ₂ ′] of Figure 4b can be obtained, and the distance between the eigenvectors of the two maps is calculated, and the Compare with the shift distance threshold to judge whether they match.

If the matching is successful, the SOH and SOC states corresponding to the successfully matched spectra are used as the SOH and SOC states of the battery to be evaluated;

If the matching fails, measure the SOH of the battery to be evaluated and the multidimensional EIS curve clusters under different SOC and temperature T conditions according to the process of step S1, and add them to the sample battery library;

S3. According to the process of step 2, match all the batteries to be evaluated;

Specifically, the matching process of all batteries to be evaluated is shown in Figure 5. Regardless of whether the current battery is successfully matched, the rest of the batteries will continue to be matched with the spectra in the spectrum library. The sample battery spectrum library is continuously improved during the evaluation process.

For a new unmatched battery in the process of obtaining the SOH of the unmatched battery and the multidimensional EIS curve cluster under different SOC and temperature T, first judge whether it matches the spectrum of the battery being tested, and if it does not match, also obtain the The SOH of the battery and the multi-dimensional EIS curve clusters under different SOC and temperature T are added to the sample spectrum library. If they match, this process is not necessary. Here it is considered that the number of unmatched batteries is small, otherwise the consistency of the whole batch of batteries to be tested is too poor, and the recycling value will be lost.

S4. Use SOH as the first indicator of consistency evaluation, set the value range of SOH, and group all the batteries to be evaluated within the value range; the value range of SOH is required by the evaluation system The accuracy is determined; for example, after selecting the value of SOH, the setting error is within the range of ±ΔSOH;

S5. Number the batteries to be evaluated in the same group, and use No. 1 battery to be evaluated as a reference battery, and calculate the similarity between the fingerprints of other batteries to be evaluated and the fingerprints of the reference battery; the calculation of the similarity of the fingerprints is still Adopt minutiae-based pattern fingerprint recognition algorithm or texture-based pattern fingerprint recognition algorithm.

S6. Taking the similarity as the second index of the consistency evaluation, judge the battery to be evaluated whose similarity is higher than the set threshold as the battery with consistency, and classify it into one category, and the remaining batteries to be evaluated follow step S5 And the process of step S6 continues to sort until all the batteries to be evaluated are sorted.

If the minutiae-based fingerprint recognition algorithm is adopted, the retired battery corresponding to the fingerprint can be considered to be consistent if the number of matching points reaches the threshold required for the similarity of the spectrum, and the more the number of matching points, the higher the similarity and the higher the consistency ; Otherwise, the decommissioned battery is considered inconsistent.

If the texture-based pattern matching algorithm is used, if the distance between the feature vectors is less than the distance threshold, it can be considered that the retired battery corresponding to the fingerprint map has consistency. The smaller the distance, the higher the similarity and the higher the consistency. consistency.

The quantity threshold and distance threshold used in the evaluation process are determined by calculating the similarity between the curve clusters corresponding to the SOH, SOC, and temperature T states of the battery to be compared in the sample battery spectrum library, and using it as the threshold. The evaluation of SOH and consistency of decommissioned power batteries is completed above.

Claims (4)

1. A method for assessing decommissioned battery SOH and consistency based on multidimensional impedance spectroscopy, characterized in that it comprises: S1. Randomly select several batteries, and measure the SOH of each battery by using a typical cycle charging and discharging method under a constant temperature state, and then measure the multidimensional EIS curve cluster of each battery under different SOC states; repeat the test after changing the temperature process, so as to obtain the multi-dimensional EIS curve cluster library of all batteries at different SOC states and temperatures T to form a sample battery spectrum library; S2. Obtain a cluster of EIS curves under excitation currents of different amplitudes with the real and imaginary parts of the impedance as coordinate axes of the battery to be evaluated and use it as its spectrum, and match it with the spectrum in the sample battery spectrum library; If the matching is successful, the SOH and SOC states corresponding to the successfully matched spectra are used as the SOH and SOC states of the battery to be evaluated; If the matching fails, measure the SOH of the battery to be evaluated and the multidimensional EIS curve clusters under different SOC and temperature T conditions according to the process of step S1, and add them to the sample battery library; S3. According to the process of step 2, match all the batteries to be evaluated; S4. Taking SOH as the first index of consistency evaluation, setting the value range of SOH, and grouping all the batteries to be evaluated within the value range into one group; S5. Number the batteries to be evaluated in the same group, and use No. 1 battery to be evaluated as a reference battery, and calculate the similarity between the fingerprints of other batteries to be evaluated and the fingerprints of the reference battery; S6. Taking the similarity as the second index of the consistency evaluation, judge the battery to be evaluated whose similarity is higher than the set threshold as the battery with consistency, and classify it into one category, and the remaining batteries to be evaluated follow step S5 And the process of step S6 continues to sort until all the batteries to be evaluated are sorted. 2. The method for evaluating decommissioned battery SOH and consistency based on multidimensional impedance spectrum according to claim 1, characterized in that, the map matching method in the step S2 is based on minutiae point pattern fingerprint recognition algorithm or based on texture pattern Fingerprint recognition algorithm. 3. the method for evaluating the decommissioned battery SOH and consistency based on multidimensional impedance spectrum according to claim 2, characterized in that, the minutiae-based pattern fingerprint recognition algorithm is to combine each graph and graph in the sample battery graph library The battery map to be evaluated is converted into a respective point set composed of feature points, and the distance between each feature point in the point set converted from the battery map to be evaluated and the corresponding feature point in each map converted point set in the map library is calculated. If the distance is less than the set If the specified offset distance threshold between points indicates that the corresponding two feature points match, count the number of feature points that match the point set converted from the battery to be evaluated with the point set converted from each map in the map library, and count the number of matching points The most atlas is used as a candidate atlas, and if the number of matching points is greater than the set threshold, the candidate atlas is a successfully matched atlas; otherwise, it indicates that the matching fails. 4. The method for evaluating decommissioned battery SOH and consistency based on multidimensional impedance spectrum according to claim 2, characterized in that, the fingerprint recognition algorithm based on texture pattern is to combine each graph in the sample battery graph library and The spectrum of the battery to be evaluated is converted into a line set composed of ridges, and the dynamic bending distance between adjacent ridges in each line set is calculated as the feature vector of the spectrum, and then the eigenvector of the battery to be evaluated is calculated and The distance between the eigenvectors of each graph in the sample battery graph library, and after selecting the minimum distance, judge whether it is less than the set graph offset distance threshold, if less, the graph corresponding to the minimum distance is regarded as a successful match spectrum; otherwise, it means that the matching failed.

Images