CN113537525A

CN113537525A - Self-adaptive early warning method for fault state of battery energy storage system

Info

Publication number: CN113537525A
Application number: CN202110836171.8A
Authority: CN
Inventors: 肖先勇; 陈智凡; 汪颖; 韦凌霄; 李瑛�; 席嫣娜; 鞠力; 陶以彬; 冯鑫振; 曹天植; 易姝娴; 陈瑞; 李烜
Original assignee: Sichuan University; China Electric Power Research Institute Co Ltd CEPRI; Electric Power Research Institute of State Grid Jibei Electric Power Co Ltd; State Grid Beijing Electric Power Co Ltd
Current assignee: Sichuan University; China Electric Power Research Institute Co Ltd CEPRI; Electric Power Research Institute of State Grid Jibei Electric Power Co Ltd; State Grid Beijing Electric Power Co Ltd
Priority date: 2021-07-23
Filing date: 2021-07-23
Publication date: 2021-10-22
Anticipated expiration: 2041-07-23
Also published as: CN113537525B

Abstract

The invention discloses a fault state self-adaptive early warning method of a battery energy storage system. The NJW clustering algorithm is used to reduce the dimension of high-dimensional monitoring data. The only alternative to the original data for cluster analysis, solves the singularity problem that often occurs in traditional methods for non-convex data clustering; uses DTW to dynamically adjust the time axis of asynchronous monitoring data, and maps the two sets of monitoring data to the synchronous time axis to overcome the problem. The observation error caused by asynchronous sampling of monitoring data is solved, and the problem that the time axis of heterologous data of sampling points cannot be one-to-one corresponding to the traditional method is solved. Finally, a sliding window model is constructed to suppress the influence of a small number of outliers in the monitoring data, and clustering is carried out based on the DTM distance. Class analysis, through sparse coefficient LSR and fault threshold FT objectively selects fault clustering points, avoids over-estimation or under-estimation of fault state by traditional fault clustering method, and realizes BESS fault state adaptive early warning.

Description

Self-adaptive early warning method for fault state of battery energy storage system

Technical Field

The invention relates to the technical field of battery energy storage system detection, in particular to a fault state self-adaptive early warning method for a battery energy storage system.

Background

Most of safety problems encountered by a Battery Energy Storage System (BESS) come from a single cell level, mainly include overcharge, overdischarge, internal short circuit and external short circuit, in the conventional BESS fault early warning, a parameter estimation method realized by establishing an accurate electric heating simulation model or a threshold value limiting method realized based on an empirical estimation method only identifies abnormal data in the operation process of the BESS, and the BESS fault early warning plays a reference correction role by depending on normal data rather than ignoring observation errors and process noise in the normal battery circulation process. However, a large-capacity battery energy storage device is generally formed by connecting hundreds of battery cells in series and in parallel, the number of monitoring units which work together with a battery box is dozens, different monitoring units perform asynchronous measurement aiming at different battery boxes, large observation errors and process noises often exist among acquired heterogeneous data, sampling time axes recorded by different monitoring units are different from a battery circulation process, data are difficult to compare, and influences of the factors on BESS fault identification precision are not negligible.

The high-capacity BESS generally divides batteries into regions, and coordinates the battery boxes of each region according to different requirements to adopt different charging and discharging strategies, for example, part of the battery boxes are deeply charged and deeply discharged to meet the peak regulation and valley filling requirements of a power plant, and the other battery boxes possibly work in a shallow charging and shallow discharging mode at the same time to inhibit user harmonic waves and solve the problem of power quality; the same battery box also has completely different working modes in different stages of charging and discharging, trickle charging can be used for protecting the safety of the battery in the initial charging stage and the later charging stage, high-rate quick charging is adopted in the middle stage, and voltage and current curves in different stages have great difference. In order to achieve sufficient identification accuracy, the fault identification method based on the measured data of the monitoring unit must jump out of a laboratory environment, and the problem that asynchronous data of battery boxes of different battery partitions are not matched in different cycle stages in actual work is solved.

Firstly, when the BESS fault state early warning is carried out by the existing method, observation errors and process noises of different source data of different battery box monitoring units are generally directly ignored or processed according to the same source noise, and no literature is provided for deeply researching a matching method of the different source data. Actual BMS (Battery Management System) monitoring data and theoretical research show that the time axes of data sequences recorded by different monitoring units are often asynchronous and difficult to directly compare, so that misjudgment is likely to occur in the fault early warning if mismatching of heterogeneous data is ignored during BESS fault state early warning.

Secondly, in the prior art, when fault identification is realized based on actually measured data of the monitoring unit, a charging and discharging strategy and a battery circulation form are generally directly defined according to the type of a battery. The high-capacity BESS used in the industrial process generally adopts different charging and discharging strategies in battery boxes of each subarea, and the same battery box also has completely different working modes in different stages of charging and discharging. If the characteristic extraction and time axis regulation are not carried out on the monitoring data, the problem that asynchronous data of different battery partition battery boxes in different cycle stages are not matched in actual work is solved, underestimation or overestimation of different degrees can easily occur in the fault state early warning, and the BESS fault state early warning cannot reflect the real safety state of the energy storage equipment.

In addition, when the existing data is used for identifying abnormal data through fault clustering, a distance method is often adopted to directly determine fault data, however, a clustering distance threshold value of a fault point is generally given through empirical estimation, so that great contingency exists, and once a clustering point set is defined too large or too small, false alarm of fault early warning is easily caused.

Interpretation of terms:

a battery energy storage system: the battery is used as an energy storage carrier and is formed by combining hundreds of battery cell monomers, and the energy storage system for storing electric energy and supplying the electric energy generally comprises a battery cabinet for storing the energy and a control cabinet for monitoring, regulating and controlling the energy.

A battery management system: the battery management system is a system for monitoring and controlling the safety and the running state of the energy storage battery, saves and processes the monitored battery information and feeds the monitored battery information back to a user in real time, and each parameter is regulated and controlled according to the acquired information to protect the battery from running reliably and stably.

NJW spectral clustering algorithm: a spectral clustering algorithm acquires a corresponding Laplacian matrix by monitoring a data similarity matrix, selects eigenvectors corresponding to a plurality of previous maximum eigenvalues as a one-to-one corresponding substitution matrix of original data, and then clusters the eigenvectors according to rows.

Dynamic time warping algorithm (DTW): an algorithm compares asynchronous time series similarity by curving a monitoring data time axis and dynamically integrates asynchronous time series.

Disclosure of Invention

Aiming at the problems, the invention aims to provide a self-adaptive early warning method for the fault state of a battery energy storage system, which is used for processing heterogeneous data based on an NJW spectral clustering algorithm, classifying similar working conditions in battery boxes capable of being compared with the fault state, solving the problem of inconsistent time axes of asynchronous data by using a dynamic warping algorithm, constructing a sliding window model to inhibit the influence of a small number of outliers in monitoring data, carrying out clustering analysis based on DTW distance, and objectively selecting fault clustering points through sparse coefficients and fault thresholds so as to realize BESS fault state self-adaptive early warning. The technical scheme is as follows:

a self-adaptive early warning method for a fault state of a battery energy storage system comprises the following steps:

step 1: reducing the dimension of high-dimensional battery box monitoring data by using an NJW clustering algorithm, obtaining a safety characteristic standard matrix by constructing a pull-type matrix of heterogeneous data, and performing clustering analysis by only replacing original data of target monitoring parameters of the battery box with the safety characteristic standard matrix;

step 2: the time axes of the safety characteristic standard matrix are regulated through a dynamic time regulation algorithm, and two groups of monitoring data are mapped to a synchronous time axis to compare the similarity degree of the asynchronous monitoring data;

and step 3: and constructing a sliding window model to inhibit the influence of a small number of outliers in the monitoring data, carrying out cluster analysis based on the DTW distance, objectively selecting fault cluster points through a sparse coefficient LSR and a fault threshold FT, and further determining the fault battery box.

Further, the step 1 specifically includes the following steps:

step 1.1: setting a target monitoring parameter of a battery management system monitoring battery box as V, acquiring m groups of battery box monitoring data, wherein each group of data comprises n sampling points, and expressing the to-be-clustered monitoring data as follows:

U＝{u₁,u₂,…,u_n}∈V^m

step 1.2: extracting battery box monitoring data to construct similarity matrix W ═ W_ij∣i≤m,j≤n}∈V^n×nThe following were used:

wherein u is_iAnd u_jRepresenting two heterogeneous battery box target monitoring data; sigma_iAnd σ_jTo adaptively identify parameters, [ sigma ]_iIs a cell box target monitoring parameter u_iAverage value of r sampling points with minimum Euclidean distance in the rest monitoring data, sigma_jIs a cell box target monitoring parameter u_jAverage value of r sampling points with minimum Euclidean distance in the rest monitoring data;

step 1.3: calculating to obtain a measurement matrix D ═ { D ═ according to the similarity matrix W_ij∣i≤m,j≤n}∈V^n×n：

Step 1.4: calculating a Laplace matrix L by the similarity matrix W and the measurement matrix D:

step 1.5: calculating the eigenvalue and eigenvector corresponding to the pull-type matrix L, and arranging the characteristics in descending orderValue and eigenvector, and taking the first K eigenvectors to form a security feature matrix S ═ S₁,s₂,…,s_K]∈V^n×KIn the method, the security feature matrix S is normalized line by line to form a security feature standard matrix Y ═ Y_ij∣i≤m,j≤n}∈V^n×K：

In the formula, s_ijIs the ith row and j columns of elements of the S matrix;

step 1.6: and each row of the full-characteristic standard matrix Y corresponds to a battery box target monitoring parameter sequence and uniquely replaces original sampling data.

Further, the step 2 specifically includes the following steps:

step 2.1: in a safety characteristic standard matrix Y, the characteristic vectors of the monitoring data corresponding to two asynchronous battery boxes a and b are respectively Y_a＝{y_a1,y_a2,…,y_anAnd Y_b＝{y_b1,y_b2,…,y_bnN is the number of sample points in each row of the safety characteristic standard matrix, namely the number of sampling points contained in the monitoring data of each battery box, and the distance matrix R of the battery boxes a and b is obtained by solving the distance matrix R ═ R_ij∣i≤n,j≤n}：

MIN＝min{r_i-1,j,r_i,j-1,r_i-1,j-1}

In the formula, d (y)_ai,y_bj) Is a sample point y_aiAnd y_bjThe Euclidean distance of; d (y)_ai,y_bj) + MIN is the sum of the minimum Euclidean distance between the current sample point and each adjacent sample point;

step 2.2: after forming the distance matrix, the DTW distance of the battery boxes a, b is:

DTW(Y_a,Y_b)＝r_nn a,b≤m

in the same way, the DTW distance of any two battery boxes measured according to the monitoring parameter V can be calculated, and when a is equal to b, the DTW distance is zero.

Further, the step 3 specifically includes the following steps:

step 3.1: sliding window of length d, d<<n, placing the initial position t of the monitoring data U to be clustered₁Moving the sliding increment backwards continuously along with the calculation of the NJW-DTW clustering algorithm to obtain sampling interval time q; calculating and storing the distance between the a-th battery box and the b-th battery box in each window until an n-d + 1-th subsequence with the length of d is formed; and so on, obtaining n-d +1 sliding windows and corresponding distance matrixes in total, and expressing as:

DTW_j＝{dtw₁,dtw₂,…,dtw_n-d+1}∈V^m

the fault condition occurs in one or more of the time windows;

step 3.2: computing sparse coefficients LSR

Defining the sparse coefficient LSR (a) of the a-th battery box as the reciprocal of the average distance of the battery box within the remaining distance k:

in the formula, size (DTW)_k(Y_a) The number of the battery boxes within the distance k is defined as the number of the battery boxes a; DTW (Y)_a,Y_b) The actual distance between the battery boxes within the k distance between the battery boxes a and b; without loss of generality, the k distance is taken to be half of the farthest DTW distance;

step 3.3: the failure threshold FT is defined as the inverse of the average distance of all battery compartments within its k distance:

step 3.4: and arranging the sparse coefficients LSR (a) of all the m battery boxes, and taking the fault threshold value FT as a judgment basis, and identifying all the battery boxes with the sparse coefficients lower than FT as fault battery boxes.

Furthermore, the battery box target monitoring parameter is battery box terminal voltage, branch current, battery temperature, insulation resistance or battery capacity.

The invention has the beneficial effects that:

1) the invention applies NJW clustering algorithm to reduce the dimension of high-dimensional monitoring data, obtains the characteristic vector of the high-dimensional monitoring data by constructing a pull-type matrix of heterogeneous data, and uniquely replaces the original data with the characteristic vector to perform clustering analysis. The NJW spectral clustering overcomes the limitation of the traditional method on the data dimension and the sequence length, and solves the singularity problem of the traditional method on the non-convex data clustering;

2) according to the method, the DTW is used for dynamically regulating the time axis of the asynchronous monitoring data, and two groups of monitoring data are mapped to the synchronous time axis, so that the observation error caused by asynchronous sampling of the monitoring data is overcome, the accuracy of a clustering result is greatly improved, and the problem that the time axes of the heterogeneous data of sampling points cannot be in one-to-one correspondence in the traditional method is solved;

3) the method constructs a sliding window model to inhibit the influence of a small number of outliers in the monitoring data, carries out clustering analysis based on the DTW distance, and objectively selects the fault clustering points through the sparse coefficient LSR and the fault threshold FT, thereby avoiding over-estimation or under-estimation of the fault state by the traditional fault clustering method, being capable of automatically adapting to parameters and actual working conditions of different BESS and realizing the self-adaptive alarm of the BESS fault state.

Drawings

Fig. 1 is a flow chart of a self-adaptive early warning method for a fault state of a battery energy storage system according to the invention.

FIG. 2 is a circular curve of the heterogeneous data on the same time axis.

Fig. 3 is a schematic view of a sliding window model.

Fig. 4 is a schematic diagram of fault clustering.

Detailed Description

The invention is described in further detail below with reference to the figures and specific embodiments. The technical scheme of the invention mainly comprises three major steps, namely heterogeneous data clustering, asynchronous sequence dynamic normalization and fault state identification, and a flow chart is shown in figure 1, wherein each major step and the minor steps thereof are elaborated as follows:

NJW spectral clustering algorithm for heterogeneous data

The NJW spectral clustering algorithm is an algorithm for carrying out dimensionality reduction clustering on high-dimensional heterogeneous data, and obtains characteristic vectors of the heterogeneous data by constructing a pull-type matrix of the heterogeneous data, and then uniquely replaces original data with the characteristic vectors to carry out clustering analysis. NJW spectral clustering has no limit on data dimensionality, and the singularity problem which often occurs in heterogenous non-convex shapes is effectively avoided. In this embodiment, the fault state early warning based on the battery box terminal voltage is taken as an example, and the steps are as follows (the rest monitoring data also have completely consistent calculation steps):

1. setting the voltage of a battery box terminal monitored by a battery management system as V, acquiring m groups of battery box monitoring data, wherein each group of data comprises n sampling points, and representing the to-be-clustered monitoring data as follows:

U＝{u₁,u₂,…,u_n}∈V^m (1)

2. extracting monitoring data to construct similarity matrix W ═ W_ij∣i≤m,j≤n}∈V^n×nThe following were used:

wherein: u. of_iAnd u_jThe terminal voltage of the battery box with two different sources is shown, and sigma is an adaptive identification parameter. Sigma_iIs a cell box target monitoring parameter u_iAverage value of r sampling points with minimum Euclidean distance in the rest monitoring data, sigma_jIs a cell box target monitoring parameter u_jAnd taking the average value of r sampling points with the minimum Euclidean distance in the rest monitoring data, wherein r is generally 3-7 in order to ensure the clustering accuracy. .

3. According to the similarityThe degree matrix W is calculated to obtain a measurement matrix D ═ D_ij∣i≤m,j≤n}∈V^n×n：

4. Calculating a Laplace matrix L by the similarity matrix W and the measurement matrix D:

5. calculating the eigenvalue and eigenvector corresponding to the pull-type matrix L, arranging the eigenvalue and eigenvector in descending order, and taking the first K eigenvectors to form a security eigenvector matrix S ═ S₁,s₂,…,s_K]∈V^n×KIn the method, S array is normalized line by line to form security feature standard matrix Y ═ Y_ij∣i≤m,j≤n}∈V^n×K：

In the formula, s_ijIs the ith row and j columns of elements of the S matrix.

6. Each row of the safety characteristic standard matrix Y corresponds to a battery box terminal voltage sequence and can uniquely replace original sampling data. And then performing dynamic time-warping clustering on the obtained data.

Dynamic time warping algorithm

As shown in fig. 2, there is a problem that the time axis of the obtained asynchronous monitoring data of different battery compartment boxes in different cycle stages in actual operation is not matched. The method improves a K mean value clustering algorithm used by a heterogeneous data NJW spectral clustering algorithm, and improves Euclidean distance into DTW distance for K mean value clustering analysis. The DTW distance compares the similarity of asynchronous monitoring data through the time axis of a regular safety feature standard matrix Y, and the steps are as follows:

1. setting two asynchronous battery boxes a and b in a safety characteristic standard matrix Y to be correspondingly monitoredThe measured data feature vectors are respectively Y_a＝{y_a1,y_a2,…,y_anAnd Y_b＝{y_b1,y_b2,…,y_bnN is the number of sample points in each row of the safety characteristic standard matrix, that is, the number of sample points included in the monitoring data of each group of battery boxes, and the distance matrix R of the battery boxes a and b can be obtained by solving the distance matrix R ═ R_ij∣i≤n,j≤n}：

MIN＝min{r_i-1,j,r_i,j-1,r_i-1,j-1} (7)

2. after forming the distance matrix, the DTW distance of the battery boxes a, b is

DTW(Y_a,Y_b)＝r_nn a,b≤m (8)

On the basis of an NJW safety characteristic standard matrix Y, the DTW method dynamically normalizes the time axis of one sequence by finding the numerical similarity of two groups of monitoring data, avoids observation errors caused by data asynchronous recording, and can map the two sequences onto a synchronous time axis, so that the similarity of the two sequences is calculated, the accuracy of a clustering result is greatly improved, and the defect that the time axes of different sampling point data cannot correspond to one another in the traditional method is overcome.

Third, fault state recognition algorithm

In order to find out abnormal fault data in monitoring data, the invention provides a fault state identification method based on an NJW-DTW clustering algorithm, which comprises the following steps, wherein in order to simplify the description, the following distance refers to a DTW distance:

1. sliding window model

And inhibiting the influence of a small number of outliers in the monitoring data by adopting a sliding window model. BMS monitoring battery box terminal voltage V acquires m groups of battery box monitoring data, and every group of data includes n sampling points, can show the monitoring data of waiting to cluster as: u ═ U₁,u₂,…,u_n}∈V^mT is more than or equal to 1 and less than or equal to n, and the length d (d)<<n) is arranged at the initial position t of the monitoring data U to be clustered₁And moving backward with the computation of the NJW-DTW clustering algorithm, the sliding increment is the sampling interval time q, as shown in FIG. 3. Calculating and storing the distance between the a-th battery box and the b-th battery box in each window until an n-d + 1-th subsequence with the length of d is formed; by analogy, n-d +1 sliding windows and corresponding distance matrixes can be obtained in total, and are represented as follows: DTW_j＝{dtw₁,dtw₂,…,dtw_n-d+1}∈V^mThe fault condition occurs in one or more of the time windows.

2. Sparse coefficient LSR (LSR)

the sparse coefficient LSR (i) reflects the distribution density of the safety states of all the battery boxes around the battery box i, and the smaller the local sparse coefficient is, the higher the failure probability of the battery box i is, and vice versa.

The failure threshold FT (FT) is defined as the inverse of the average distance of all battery compartments within its k distance:

if a battery box i fails, its sparseness factor should be smaller than the failure factor FT, because the battery box i is now located a greater safety distance from all other battery boxes than all other battery boxes. All the battery boxes with sparse coefficients LSR smaller than the fault factor FT are taken as candidate fault battery boxes, and fault clustering is shown in FIG. 4.

3. Fault state identification

And (3) arranging the sparse coefficients LSR (a) of all the m battery boxes, and taking the fault threshold value FT as a judgment basis, wherein all the battery boxes with the sparse coefficients lower than FT can be regarded as fault battery boxes.

Starting from heterogeneous monitoring data, the method applies an NJW clustering algorithm to reduce the dimension of high-dimensional data, extracts the characteristic vector of the monitoring data, and eliminates observation errors and process noise influence; the DTW method is used for dynamically integrating the asynchronous data, so that the problem of mismatching of heterogeneous data is solved; the method has the advantages that fault clustering analysis is carried out based on objective indexes, and the method has strong self-adaptability to different systems, so that the safe operation state of the battery energy storage system can be objectively reflected, and fault early warning can be accurately sent out.

Claims

1. A battery energy storage system fault state adaptive early warning method, characterized in that, comprising the following steps:

Step 1: Use the NJW clustering algorithm to reduce the dimension of the high-dimensional battery box monitoring data, obtain its safety feature standard matrix by constructing a pull matrix of heterogeneous data, and then use the safety feature standard matrix to uniquely replace the battery box target monitoring parameters Cluster analysis of raw data;

Step 2: Regularize the time axis of the security feature standard matrix through the dynamic time warping algorithm, and map the two sets of monitoring data to the synchronous time axis to compare the similarity of the asynchronous monitoring data;

Step 3: Build a sliding window model to suppress the influence of a small number of outliers in the monitoring data, perform cluster analysis based on the DTW distance, objectively select the fault clustering points through the sparse coefficient LSR and the fault threshold FT, and then determine the faulty battery box.

2. The battery energy storage system fault state adaptive early warning method according to claim 1, wherein the step 1 specifically comprises the following steps:

Step 1.1: Set the target monitoring parameter of the battery management system to monitor the battery box as V, and obtain m groups of battery box monitoring data, each group of data contains n sampling points, and the monitoring data to be clustered is expressed as:

U={u ₁ ,u ₂ ,…,u _n }∈V ^m

Step 1.2: Extract the battery box monitoring data and construct a similarity matrix W={w _ij ∣i≤m,j≤n}∈V ^n×n as follows:

Among them, u _i and u _j represent two heterogeneous battery box target monitoring data; σ _i and σ _j are adaptive identification parameters, σ _i is the battery box target monitoring parameter u _i and the remaining monitoring data in the smallest Euclidean distance r The average value of the sampling points, σ _j is the average value of the r sampling points with the smallest Euclidean distance between the target monitoring parameter u _j of the battery box and the rest of the monitoring data;

Step 1.3: Calculate the metric matrix D={d _ij ∣i≤m,j≤n}∈V ^n×n according to the similarity matrix W:

Step 1.4: Calculate the Laplace matrix L from the similarity matrix W and the metric matrix D:

Step 1.5: Calculate the eigenvalues and eigenvectors corresponding to the pull matrix L, arrange the eigenvalues and eigenvectors in descending order, and take the first K eigenvectors to form a security feature matrix S=[s ₁ ,s ₂ ,...,s _K ]∈ In V ^n×K , the security feature matrix S is normalized row by row to form the security feature standard matrix Y={y _ij ∣i≤m,j≤n}∈V ^n×K :

In the formula, s _ij is the element of the ith row and j column of the S matrix;

Step 1.6: Each row of the full-feature standard matrix Y corresponds to a battery box target monitoring parameter sequence, which uniquely replaces the original sampling data.

3. The battery energy storage system fault state adaptive early warning method according to claim 2, wherein the step 2 specifically comprises the following steps:

Step 2.1: In the safety feature standard matrix Y, the monitoring data feature vectors corresponding to the two asynchronous battery boxes a and b are respectively Y _a ={y _a1 ,y _a2 ,...,y _an } and Y _b ={y _b1 , _y _b2 , . _ij ∣i≤n,j≤n}:

MIN=min{r _i-1,j ,r _i,j-1 ,r _i-1,j-1 }

In the formula, d(y _ai , y _bj ) is the Euclidean distance between the sample points y _ai and y _bj ; d(y _ai , y _bj )+MIN is the sum of the minimum Euclidean distances between the current sample point and the neighboring sample points;

Step 2.2: After forming the distance matrix, the DTW distances of battery boxes a and b are:

DTW(Y _a ,Y _b )=r _nn a,b≤m

In the same way, the DTW distance measured by any two battery boxes according to the monitoring parameter V can be calculated. When a=b, the DTW distance is zero.

4. The battery energy storage system fault state adaptive early warning method according to claim 3, wherein the step 3 specifically comprises the following steps:

Step 3.1: Place the sliding window of length d, d<<n, at the starting position t ₁ of the monitoring data U to be clustered, and move backward with the calculation of the NJW-DTW clustering algorithm. The sliding increment is Sampling interval time q; calculate and save the distances of the a and b battery boxes in each window until the n-d+1 subsequence of length d is formed; and so on, a total of n-d+1 are obtained The sliding window and the corresponding distance matrix are expressed as:

DTW _j ={dtw ₁ ,dtw ₂ ,...,dtw _n-d+1 }∈V ^m

The fault condition occurred in one or more of these time windows;

Step 3.2: Calculate the sparse coefficient LSR

The sparsity coefficient LSR(a) of the a-th battery box is defined as the reciprocal of the average distance between this battery box and the remaining battery boxes within the distance k:

In the formula, size (DTW _k (Y _a )) is the number of battery boxes within the distance k of battery box a; DTW (Y _a , Y _b ) is the actual distance of battery boxes a and b within the distance k; no In general, the k distance is taken as half of the farthest DTW distance;

Step 3.3: Define the fault threshold FT as the reciprocal of the average distance of all battery boxes to the battery boxes within k distances:

Step 3.4: Arrange the sparse coefficients LSR(a) of all m battery boxes, and use the fault threshold FT as the judgment basis, and identify all battery boxes with sparse coefficients lower than FT as faulty battery boxes.

5. The battery energy storage system fault state adaptive early warning method according to any one of claims 1-4, wherein the target monitoring parameters of the battery box are battery box terminal voltage, branch current, battery temperature, insulation resistance or battery capacity.