New energy storage battery inconsistency evaluation method fusing Hilbert-Huang transform and information entropy

Abstract

The invention belongs to the field of energy storage battery management systems, and discloses a novel energy storage battery inconsistency evaluation method fusing Hilbert-Huang transform and information entropy. According to the method, a series of IMF (intrinsic mode function) and RES (residual error component) after different state quantities of the energy storage battery cell are subjected to EMD (empirical mode decomposition) and HHT (Hilbert Huang Transform) are fused to serve as evaluation factors, the inconsistency of the energy storage battery cell is subjected to frequency analysis, and finally, the fused evaluation factors serve as evaluation factors. Calculating the information entropy of each energy storage battery cell, and describing the consistency level of each energy storage battery cell through the information entropy; and finally, performing feature fusion by using a cluster attention mechanism, and evaluating the inconsistency level of the battery cluster by taking a result after feature fusion as an information entropy score value. The method provided by the invention is not limited by the number of the battery cells and the type and number of evaluation factors, and has wide adaptability.

Description

Technical Field

[0001] The invention belongs to the field of energy storage battery management systems and relates to a new method for evaluating inconsistency of energy storage batteries by integrating Hilbert-Huang transform and information entropy. Background Art

[0002] New energy storage technologies have excellent control characteristics, especially high energy density, fast response speed, high conversion efficiency and short construction period. The application scenarios of energy storage power stations are very broad, but in the application process of electrochemical energy storage, due to the consistency of batteries, control system errors and lack of operation and maintenance detection methods, the availability of electrochemical energy storage systems is reduced, the failure rate is frequent, and even serious fires occur.

[0003] The development of energy storage technology is one of the key ways to solve environmental and energy problems. Before the battery leaves the factory, it must be filtered according to its performance and then connected in series and parallel to the battery pack. The manufacturing process of energy storage batteries is complex, including electrode manufacturing, assembly, forming and testing processes, and there are inevitably production and material inhomogeneities. Specifically, the electrode thickness, electrolyte concentration and SEI film formation may vary from cell to cell. Even on the same production line and the same batch, parameters such as capacity, internal resistance and self-discharge rate may be inconsistent. This can be called "innate" inconsistency. On the other hand, due to the complex charging and discharging environment, there are differences in temperature and ventilation between cells, which worsens the "innate" inconsistency and forms the "acquired environment" inconsistency. This inconsistency will cause the "short board effect" of the battery and shorten the battery life.

[0004] The energy storage battery system is the core of the energy storage power station, which is generally composed of hundreds or thousands of batteries connected in series and parallel. Due to the slight deviations and uncertainties in the production process of the battery cells, the battery cells have inherent inconsistencies. And they work in a complex charging and discharging environment for a long time, so their inconsistency will be further aggravated, which greatly reduces the available capacity and shortens the service life. Therefore, it is urgent to establish a battery inconsistency evaluation mechanism to provide a basis for efficient and reliable management. The various reactions and aging mechanisms inside the energy storage battery are complex, and the existing research methods have certain shortcomings. Wang Xiaohui proposed a battery inconsistency evaluation method based on voltage extreme difference in "Research on Balanced Control Strategy of Power Battery Pack". Since the working voltage of the battery will be affected by state quantities such as current and temperature, the evaluation method based on working voltage is not accurate; Jin Weiren et al. proposed a battery inconsistency evaluation method based on battery state of charge (SOC) standard deviation and extreme difference in "Research Progress of Lithium-ion Power Battery Consistency Evaluation Method". Compared with the method of using working voltage evaluation alone, the method of using SOC evaluation is more reasonable. However, since SOC is not a battery state quantity that can be directly measured, the error of SOC estimation itself will lead to further amplification of the error of evaluating battery inconsistency, which makes this evaluation method based on SOC also subject to certain limitations; Liye Wang et al. in "Research on Multi-Parameter Evaluation of Electric Vehicle Power Battery Consistency Based on Principal Component In the article "Consistency evaluation and cluster analysis for lithium-ion battery pack in electric vehicles", the battery consistency characteristic parameters were analyzed, and a multi-parameter evaluation function for battery consistency was established, which solved the problem of using a single parameter to evaluate battery consistency in battery balancing technology. However, this method cannot be generally applied to different types of batteries, which makes this method also limited; Jiaqiang Tian et al. proposed an online consistency evaluation method for lithium-ion battery packs based on multi-feature weighting in "Consistency evaluation and cluster analysis for lithium-ion battery pack in electric vehicles", which simultaneously considered open circuit voltage, ohmic internal resistance and polarization internal resistance, and improved the accuracy of consistency evaluation. However, since the batteries of energy storage power stations cannot be disassembled in the working state, the ohmic internal resistance and polarization internal resistance cannot be measured, so this method is also difficult to apply to the actual working conditions of energy storage power stations; battery inconsistency is often a comprehensive characteristic of the battery in terms of capacity, internal resistance, self-discharge rate, etc., and the single parameter evaluation method cannot fully reflect the true situation of battery inconsistency. Therefore, the multi-parameter evaluation method is appropriate.However, the existing evaluation methods are often simple mathematical operations with poor reliability. Considering the universal applicability of the evaluation methods and the actual working conditions of energy storage power stations, in order to improve the safety performance of energy storage power stations, reduce the potential safety hazards of energy storage power stations, and ensure that the batteries in energy storage power stations emit normal power, a new method for analyzing the consistency level of energy storage batteries needs to be proposed. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention proposes a comprehensive evaluation method for inconsistency of energy storage batteries with high reliability and good adaptability.

[0006] The technical solution of the present invention:

[0007] A new method for evaluating the inconsistency of energy storage batteries by integrating Hilbert Huang transform and information entropy is proposed. The HHT transform is combined with information entropy to evaluate the inconsistency level of energy storage battery cells and battery clusters. A series of intrinsic mode functions (IMF) and residual components (RES) of different state quantities of energy storage battery cells after empirical mode decomposition (EMD) and HHT transform are fused as evaluation factors to perform frequency analysis on the inconsistency of energy storage battery cells. Finally, the fused evaluation factors are used as evaluation factors to calculate the information entropy of each energy storage battery cell, and the consistency level of each energy storage battery cell is described by information entropy. Finally, the cluster attention mechanism is used for feature fusion, and the result of feature fusion is used as the information entropy score value to evaluate the inconsistency level of the battery cluster. The specific steps are as follows:

[0008] (1) BMS data collection

[0009] The battery management system collects energy storage battery data to obtain different state quantities of the energy storage battery cells, including voltage and temperature;

[0010] (2) Perform EMD decomposition and HHT transformation on different state quantities, and finally use information entropy to evaluate the inconsistency of energy storage batteries;

[0011] The steps of EMD decomposition and HHT transformation of different state quantities of energy storage battery cells are as follows:

[0012] 1) First, for one energy storage battery cell, there are N different state quantities. EMD decomposition and HHT transformation are performed on the N different state quantities. One state quantity is decomposed to obtain one RES and n1 IMFs, thereby obtaining n2 RES of N different state quantities and n3 IMFs of these N different state quantities; among which, n2 = 1*N, n3 = n1*N; then the weighted average of the n2 residual components is performed to obtain a fused RES after the fusion of one energy storage battery cell:

[0013]

[0014] In the formula, r i is the i-th RES of N different state quantities, i=1,2,…,n2; n2 is the total number of RES corresponding to N different state quantities; r is the fused RES after the fusion of n2 RES of N different state quantities corresponding to one energy storage battery cell;

[0015] 2) For the total n3 IMFs of the N different state quantities obtained above, the sum C of the total n1 IMFs corresponding to the kth state quantity is k As a component of a set, the set is composed as follows:

[0016]

[0017] Where C1, C2, …, C k is the sum of the total n1 IMFs corresponding to the kth state quantity corresponding to one energy storage battery cell, n1 is the number of IMFs decomposed from each state quantity, c ik is the ikth IMF corresponding to the kth state quantity, ik=1,2,…,n1, k=1,2,…,N;

[0018] 3) During the fusion process, the weighted coefficient is calculated through the following relationship:

[0019]

[0020] In the formula, σ ik is the standard deviation of the n1 IMFs corresponding to the state quantity of the kth energy storage battery cell, w k is the weighted coefficient corresponding to the state quantity of the kth energy storage battery cell;

[0021] Thus, the weighted coefficient w corresponding to the state quantity of the kth energy storage battery cell is obtained k , according to this weighted coefficient, the sum of the n1 IMFs corresponding to each state quantity is fused:

[0022]

[0023] In the formula, C k is the sum of the total n1 IMFs corresponding to the kth state quantity corresponding to one energy storage battery cell, w K is the weighted coefficient corresponding to the state quantity of the kth energy storage battery cell, and C0 is the final fusion IMF of one cell;

[0024] The final fusion result is the sum of the fused IMF C0 and the fused RES r:

[0025] y m =C0+r

[0026] In the formula, y m represents the final fusion evaluation factor of the mth energy storage battery cell, m = 1, 2, ..., n, and n is the total number of evaluated energy storage battery cells;

[0027] The fusion result y m As the input evaluation factor for further calculating the information entropy of the mth energy storage battery cell;

[0028] (3) Battery inconsistency analysis

[0029] Fusion result y m The information entropy H m It is expressed as:

[0030] H m =-y m lg m ,m=1,2,...,n

[0031] In the formula, H m Represents the information entropy of the mth energy storage battery cell;

[0032] Select n energy storage battery cells as the evaluation object and select the final fusion factor y after fusion m As the status evaluation factor of the mth energy storage battery cell on that day; calculate the information entropy of the mth energy storage battery cell respectively, and the calculation steps are as follows:

[0033] 1) Assuming that the number of energy storage battery cells in the battery cluster is n, the original data matrix Y0 of the evaluation model is:

[0034] Y0=(y1,y2,...,y m ),m=1,2,...,n

[0035] 2) Normalize the original data matrix Y0 to obtain the normalized matrix Y1 of all evaluation factors:

[0036] Y1=(v1,v2,...,v m ),m=1,2,...,n

[0037] In the formula, v m Yes m The normalized weight;

[0038] 3) The information entropy of the mth energy storage battery cell is as follows:

[0039] H m =-v m lgv m ,m=1,2,...,n

[0040] 4) After obtaining the information entropy of each energy storage battery cell, the cluster attention mechanism network is used to fuse the information entropy of the n energy storage battery cells. The fusion process is as follows: first, the information entropy H1 and the information entropy H2 are fused using the cluster attention mechanism network to obtain a fusion factor 1, and then the fusion factor 1 is fused with the information entropy H3 using the cluster attention mechanism network to obtain a fusion factor 2, and so on to obtain a fusion factor n-1. The fusion factor n-1 is used as an information entropy score evaluation value Z of the final cluster of n energy storage battery cells. The information entropy score evaluation value Z is calculated every fixed time. The information entropy H of each cell within a day is used to obtain the information entropy score evaluation value Z. m Qualitatively analyze the energy storage battery cells with the worst consistency level; describe the degree of deviation of the information entropy of a cluster of energy storage battery cells based on the information entropy score evaluation value Z of a cluster of energy storage battery cells over a period of time, and determine the inconsistency of the entire battery cluster under different life states.

[0041] Beneficial effects of the present invention: The present invention uses directly measured physical quantities of energy storage batteries as input, which reduces the complexity of data processing. It introduces the Hilbert-Huang transform and fuses information entropy to calculate and evaluate the consistency of battery PACK and battery cluster. Experimental results show that this method can scientifically evaluate the inconsistency of battery PACK and battery cluster at different lifespans, and has universal applicability. In particular, this method is not limited by the number of battery cells, the type and number of evaluation factors. Therefore, this energy storage battery inconsistency evaluation method that combines Hilbert-Huang transform and information entropy can be used as an effective method for evaluating battery inconsistency and has wide adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 For this method, an inconsistency analysis flow chart is performed based on the battery of the energy storage power station;

[0043] Figure 2 A technical roadmap for evaluating battery inconsistency by integrating HHT and information entropy;

[0044] Figure 3 Schematic diagram of cluster attention mechanism. DETAILED DESCRIPTION

[0045] The specific implementation of the present invention is further described below in conjunction with the accompanying drawings and technical solutions.

[0046] Example

[0047] A new method for evaluating inconsistency of energy storage batteries by integrating Hilbert-Huang transform and information entropy is proposed. The specific steps are as follows:

[0048] 1) If Figure 1 and Figure 2As shown in the figure, firstly, the 4 state quantity data of cell voltage, temperature, SOC and current collected by the BMS of the energy storage power station are subjected to EMD decomposition and HHT transformation, and the number of RES for each state quantity decomposition is set to 1 and the number of IMF is set to 5, thus obtaining a total of 4 RES and 20 IMF. The fused RES is obtained by weighted average fusion of these 4 RES:

[0049]

[0050] In the formula, r i is the i-th RES of 4 different state quantities, i=1,2,…,4; r is the fused RES after fusion of 4 RES of 4 different state quantities corresponding to 1 energy storage battery cell;

[0051] 2) For the total 20 IMFs of the 4 different state quantities obtained above, the sum of the 5 IMFs corresponding to the kth state quantity is C k As a component of a set, the set is composed as follows:

[0052]

[0053] Where C1, C2, …, C k is the sum of the five IMFs corresponding to the kth state quantity of a storage battery cell, c ik is the ikth IMF corresponding to the kth state quantity, ik=1,2,…,5, k=1,2,…,4;

[0054] 3) Fuse these 20 IMFs and 4 RESs to obtain the final fusion result of a battery cell.

[0055] Calculate the weighting coefficients:

[0056]

[0057] In the formula, σ ik is the standard deviation of the n1 IMFs corresponding to the state quantity of the kth energy storage battery cell, w k is the weighting coefficient corresponding to the state quantity of the k-th energy storage battery cell, k = 1, 2, ..., 4;

[0058] Thus, the weighted coefficient w corresponding to the state quantity of the kth energy storage battery cell is obtained k , k = 1, 2, ..., 4, and the sum of the five IMFs corresponding to each state quantity is fused according to this weighting coefficient:

[0059]

[0060] In the formula, C kis the sum of the five IMFs corresponding to the kth state quantity of a storage battery cell, w K is the weighted coefficient corresponding to the state quantity of the kth energy storage battery cell, k = 1, 2, ..., 4, C0 is the final fusion IMF of a cell;

[0061] The final fusion result is the sum of the fused IMF C0 and the fused RES r:

[0062] y m =C0+r

[0063] In the formula, y m represents the final fusion evaluation factor of the mth energy storage battery cell, m = 1, 2, ..., 30;

[0064] 4) Perform battery inconsistency analysis:

[0065] The original data matrix Y0 of the evaluation model is:

[0066] Y0=(y1,y2,...,y m ),m=1,2,...,30

[0067] Normalize the original data matrix Y0 to obtain the normalized matrix Y1 of all evaluation factors:

[0068] Y1=(v1,v2,...,v m ),m=1,2,...,30

[0069] In the formula, v m Yes m The normalized weight;

[0070] Calculate the information entropy of the mth energy storage battery cell:

[0071] H m =-y m lg m ,m=1,2,...,30

[0072] In the formula, H m Represents the information entropy of the mth energy storage battery cell;

[0073] After obtaining the information entropy of each energy storage battery cell, the cluster attention mechanism network is used to fuse the information entropy of these 30 energy storage battery cells, such as Figure 3As shown in Figure 1, the fusion process is as follows: first, the information entropy H1 and the information entropy H2 are fused using the cluster attention mechanism network to obtain the fusion factor 1, and then the fusion factor 1 is fused with the information entropy H3 using the cluster attention mechanism network to obtain the fusion factor 2, and so on to obtain the fusion factor 29, and the fusion factor 29 is used as the final information entropy score evaluation value Z of the entire cluster of 30 energy storage battery cells. The information entropy score evaluation value Z is calculated every fixed time, and the information entropy H of each cell within a day is used to calculate the information entropy score evaluation value Z. m Qualitative analysis of the energy storage battery cell with the worst consistency level, information entropy H m The smaller the value, the worse the consistency level of the mth battery cell is. According to the information entropy score evaluation value Z of a cluster of energy storage battery cells over a period of time, the deviation degree of the information entropy of a cluster of energy storage battery cells is described, and the inconsistency of the entire battery cluster under different life states is determined. The larger the information entropy score evaluation value Z is, the worse the consistency level of this cluster of batteries is.

[0074] The present invention and its embodiments are described schematically above, and the description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by it and designs a structural method and an embodiment similar to the technical solution without creativity without departing from the purpose of the invention, they shall all fall within the protection scope of the present invention.

Claims

1. A new method for evaluating inconsistency of energy storage batteries by integrating Hilbert Huang transform and information entropy, characterized in that: The HHT transformation is combined with information entropy to evaluate the inconsistency level of energy storage battery cells and battery clusters. A series of intrinsic mode functions IMF and residual components RES of different state quantities of energy storage battery cells after EMD decomposition and HHT transformation are fused as evaluation factors, and the inconsistency of energy storage battery cells is analyzed by frequency division. Finally, the fused evaluation factors are used as evaluation factors to calculate the information entropy of each energy storage battery cell, and the consistency level of each energy storage battery cell is described by information entropy. Finally, the cluster attention mechanism is used for feature fusion, and the result of feature fusion is used as the information entropy score value to evaluate the inconsistency level of the battery cluster. The specific steps are as follows: (1) BMS data collection The battery management system collects energy storage battery data to obtain different state quantities of the energy storage battery cells, including voltage and temperature; (2) Perform EMD decomposition and HHT transformation on different state quantities, and finally use information entropy to evaluate the inconsistency of the energy storage battery; The steps of EMD decomposition and HHT transformation of different state quantities of energy storage battery cells are as follows: 1) First, for one energy storage battery cell, there are N different state quantities. EMD decomposition and HHT transformation are performed on the N different state quantities. One state quantity is decomposed to obtain one RES and n1 IMFs, thereby obtaining n2 RES of N different state quantities and n3 IMFs of these N different state quantities; among which, n2 = 1*N, n3 = n1*N; then the weighted average of the n2 residual components is performed to obtain a fused RES after the fusion of one energy storage battery cell: In the formula, r i is the i-th RES of N different state quantities, i=1,2,…,n2; n2 is the total number of RES corresponding to N different state quantities; r is the fused RES after the fusion of n2 RES of N different state quantities corresponding to one energy storage battery cell; 2) For the total n3 IMFs of the N different state quantities obtained above, the sum C of the total n1 IMFs corresponding to the kth state quantity is k As a component of a set, the set is composed as follows: {C1,C2,...,C k },k=1,2,...,N Where C1, C2, …, C k is the sum of the total n1 IMFs corresponding to the kth state quantity corresponding to one energy storage battery cell, n1 is the number of IMFs decomposed from each state quantity, c ik is the ikth IMF corresponding to the kth state quantity, ik=1,2,…,n1, k=1,2,…,N; 3) During the fusion process, the weighted coefficient is calculated through the following relationship: In the formula, σ ik is the standard deviation of the n1 IMFs corresponding to the state quantity of the kth energy storage battery cell, w k is the weighted coefficient corresponding to the state quantity of the kth energy storage battery cell; Thus, the weighted coefficient w corresponding to the state quantity of the kth energy storage battery cell is obtained k , according to this weighted coefficient, the sum of the n1 IMFs corresponding to each state quantity is fused: In the formula, C k is the sum of the total n1 IMFs corresponding to the kth state quantity corresponding to one energy storage battery cell, w K is the weighted coefficient corresponding to the state quantity of the kth energy storage battery cell, and C0 is the final fusion IMF of one cell; The final fusion result is the sum of fused IMF C0 and fused RESr: y m =C0+r In the formula, y m represents the final fusion evaluation factor of the mth energy storage battery cell, m = 1, 2, ..., n, and n is the total number of evaluated energy storage battery cells; The fusion result y m As the input evaluation factor for further calculating the information entropy of the mth energy storage battery cell; (3) Battery inconsistency analysis Fusion result y m The information entropy H m It is expressed as: H m =-y m lgy m ,m=1,2,...,n In the formula, H m Represents the information entropy of the mth energy storage battery cell; Select n energy storage battery cells as the evaluation object and select the final fusion factor y after fusion m As the status evaluation factor of the mth energy storage battery cell on that day; calculate the information entropy of the mth energy storage battery cell respectively, and the calculation steps are as follows: 1) Assuming that the number of energy storage battery cells in the battery cluster is n, the original data matrix Y0 of the evaluation model is: Y0=(y1,y2,...,y m ),m=1,2,...,n 2) Normalize the original data matrix Y0 to obtain the normalized matrix Y1 of all evaluation factors: Y1=(v1,v2,...,v m ),m=1,2,...,n In the formula, v m Yes m The normalized weight; 3) The information entropy of the mth energy storage battery cell is as follows: H m =-v m lgv m ,m=1,2,...,n 4) After obtaining the information entropy of each energy storage battery cell, the cluster attention mechanism network is used to fuse the information entropy of the n energy storage battery cells. The fusion process is as follows: first, the information entropy H1 and the information entropy H2 are fused using the cluster attention mechanism network to obtain a fusion factor 1, and then the fusion factor 1 is fused with the information entropy H3 using the cluster attention mechanism network to obtain a fusion factor 2, and so on to obtain a fusion factor n-1. The fusion factor n-1 is used as an information entropy score evaluation value Z of the final cluster of n energy storage battery cells. The information entropy score evaluation value Z is calculated every fixed time. The information entropy H of each cell within a day is used to obtain the information entropy score evaluation value Z. m Qualitatively analyze the energy storage battery cells with the worst consistency level; describe the degree of deviation of the information entropy of a cluster of energy storage battery cells based on the information entropy score evaluation value Z of a cluster of energy storage battery cells over a period of time, and determine the inconsistency of the entire battery cluster under different life states.

Images