Battery fault monitoring method, battery management system, battery device and electric device

Abstract

The application discloses a battery fault monitoring method, a battery management system, a battery device and a power utilization equipment, and belongs to the technical field of battery management. The method comprises the following steps: acquiring a voltage matrix of a battery in a running process; performing low-rank approximation processing on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix; extracting a principal component projection feature of each battery cell from a left singular vector matrix or a right singular vector matrix of the low-rank approximation matrix; calculating a residual matrix of the voltage matrix and the low-rank approximation matrix, and extracting at least one residual feature of each battery cell from the residual matrix; weighting and fusing the residual feature and the principal component projection feature of each battery cell according to respective corresponding weights to obtain a fault monitoring feature of each battery cell; and identifying a fault battery cell in the battery according to the fault monitoring features of all the battery cells. The application can accurately identify a battery cell in an early fault stage.

Description

Technical Field

[0001] This application belongs to the field of battery management technology, and in particular relates to a battery fault monitoring method, a battery management system, a battery device, and an electrical device. Background Technology

[0002] As a core component of new energy vehicles, the safety of power batteries directly affects the life and property safety of passengers and the operational reliability of the vehicle. Thermal runaway of power batteries is uncertain and sudden, and can occur under various operating conditions, including when the vehicle is stationary, in motion, or charging.

[0003] Currently, the assessment of the safety performance of power batteries mainly relies on threshold diagnosis of intuitive parameters such as voltage, temperature, and current. Alarms are only triggered when the parameters exceed the set thresholds. This lacks effective identification of early minor anomalies and suffers from problems such as low computational efficiency and low fault identification rate, making it difficult to meet the needs of real-time monitoring and accurate early warning.

[0004] In summary, improving the monitoring of the battery status of new energy vehicles and promptly detecting potential battery faults and issuing early warnings have become important technical issues that urgently need to be addressed. Summary of the Invention

[0005] This application provides a battery fault monitoring method, a battery management system, a battery device, and an electrical appliance, which can accurately identify battery cells in the early stages of fault.

[0006] On one hand, embodiments of this application provide a battery fault monitoring method, the method comprising: Obtain the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. The voltage matrix is ​​approximated by a low-rank process to obtain a low-rank approximation matrix of the voltage matrix; Extract the principal component projection features of each battery cell from the left or right singular vector matrix of the low-rank approximation matrix; Calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature for each cell from the residual matrix; The residual features and principal component projection features of each cell are weighted and fused according to their respective weights to obtain the fault monitoring features of each cell. Based on the fault monitoring characteristics of all cells, faulty cells in the battery are identified.

[0007] On the other hand, embodiments of this application provide a battery management system, including: A sampling circuit is used to acquire the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. A control circuit is configured to perform a low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix; extract principal component projection features of each cell from the left or right singular vector matrix of the low-rank approximation matrix; calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature of each cell from the residual matrix; perform weighted fusion of the residual features of each cell and the principal component projection features according to their respective weights to obtain the fault monitoring features of each cell; and identify the faulty cells in the battery based on the fault monitoring features of all cells.

[0008] In another aspect, embodiments of this application provide a battery device, which includes a plurality of battery cells and a battery management system as described above.

[0009] In another aspect, embodiments of this application provide an electrical device, including the battery device described above.

[0010] Furthermore, embodiments of this application provide a battery fault monitoring device, which includes: The acquisition module is used to acquire the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. The processing module is used to perform low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix; The extraction module is used to extract the principal component projection features of each battery cell from the left singular vector matrix or the right singular vector matrix of the low-rank approximation matrix. The calculation module is used to calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature of each cell from the residual matrix; The fusion module is used to perform weighted fusion of the residual features and the principal component projection features of each cell according to their respective weights to obtain the fault monitoring features of each cell. An identification module is used to identify faulty cells in the battery based on the fault monitoring characteristics of all cells.

[0011] In another aspect, embodiments of this application provide an electronic device, which includes: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions stored in the memory, it implements the battery fault monitoring method provided in one aspect.

[0012] In another aspect, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the battery fault monitoring method provided in one aspect.

[0013] The battery fault monitoring method, battery management system, battery device, and electrical equipment of this application embodiment acquire the voltage matrix of the battery during operation, perform low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix, which can accurately reflect the group consistency behavior of the cells in the battery under the current operating conditions and adapt to various changing operating environments. From the left singular vector matrix or the right singular vector matrix of the low-rank approximation matrix, the principal component projection features of each cell are extracted, which can reflect the projection features of each cell in the main behavior patterns of all cells in the battery. The residual matrix of the voltage matrix and the low-rank approximation matrix is ​​calculated, and at least one residual feature of each cell is extracted from the residual matrix, which can reflect the small voltage deviation of the early fault of the cell on the group consistency. The residual features of each cell and the principal component projection features are weighted and fused according to their respective weights to obtain the fault monitoring features of each cell, which can reflect the true health status of the cell from multiple dimensions. Then, based on the fault monitoring features of all cells, the faulty cells in the battery can be identified, and cells in the early stage of fault in the battery can be accurately identified. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic flowchart of a battery fault monitoring method provided in one embodiment of this application; Figure 2 This is a flowchart illustrating a battery fault monitoring method provided in another embodiment of this application; Figure 3 This is a schematic diagram of the structure of a battery fault monitoring device provided in another embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in another embodiment of this application. Detailed Implementation

[0016] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0017] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0018] As a core component of new energy vehicles, the safety of power batteries directly affects the life and property safety of passengers and the operational reliability of the vehicle. Thermal runaway of power batteries is uncertain and sudden, and can occur under various operating conditions, including when the vehicle is stationary, in motion, or charging.

[0019] Currently, the assessment of the safety performance of power batteries mainly relies on threshold diagnosis of intuitive parameters such as voltage, temperature, and current. Alarms are only triggered when the parameters exceed the set thresholds. This lacks effective identification of early minor anomalies and suffers from problems such as low computational efficiency and low fault identification rate, making it difficult to meet the needs of real-time monitoring and accurate early warning.

[0020] In summary, improving the monitoring of the battery status of new energy vehicles and promptly detecting potential battery faults and issuing early warnings have become important technical issues that urgently need to be addressed.

[0021] Regarding the execution subject used in the technical solutions provided in the embodiments of this application, the execution subject of the battery fault monitoring method can be a battery management system, or it can be a server or terminal device for remotely controlling the battery management system, etc.

[0022] In addition, the execution entity used in the embodiments of this application can also be a software execution entity, such as a client or software program installed in the battery management system or terminal device. The specific types of execution entities corresponding to the battery fault monitoring method, battery management system, battery device, and electrical equipment provided in the embodiments of this application are not strictly limited here; they can be flexibly selected and set according to the application scenario and actual needs.

[0023] It should be noted that the specific application scenarios of the battery fault monitoring method, battery management system, battery device and electrical equipment provided in the embodiments of this application are not strictly limited. The battery fault monitoring method, battery management system, battery device and electrical equipment provided in the embodiments of this application can be flexibly applied to various application scenarios that require battery fault monitoring according to actual needs.

[0024] To address the problems of the prior art, embodiments of this application provide a battery fault monitoring method, a battery management system, a battery device, and an electrical device. The battery fault monitoring method provided in this application embodiment will be described first below.

[0025] Figure 1 This is a schematic flowchart of a battery fault monitoring method provided in one embodiment of this application. Figure 1 As shown, in one embodiment, the battery fault monitoring method includes steps 101 to 106.

[0026] Step 101: Obtain the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery.

[0027] A battery cell, also known as a single battery cell, is formed by connecting two or more cells in series, parallel, or a combination thereof.

[0028] The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. The rows and columns of the voltage matrix correspond to the sampling time and the voltage of multiple cells in the battery, respectively.

[0029] In one implementation, the rows of the voltage matrix correspond to the sampling time, the columns correspond to the individual cells in the battery, and the matrix elements are the voltage values ​​of the corresponding cells at the corresponding sampling time. This allows for real-time acquisition of voltage data from each cell during actual battery operation, forming the voltage matrix. In the voltage matrix, rows correspond to the sampling time, and columns correspond to the individual cells in the battery. The first element in the voltage matrix... The line represents the first The voltage values ​​of all cells at the sampling time, the first value in the voltage matrix. Column represents the first The voltage values ​​of each cell at all sampling times, wherei , j It is a natural number.

[0030] In one implementation, the voltage data of each cell in the battery, acquired in real time during actual operation, can be preprocessed to obtain a voltage matrix. The preprocessing may include at least one of the following: removing outliers (such as invalid or zero values), data alignment and sorting, and dividing the data by operational segments (such as the charging and discharging process).

[0031] For example, the voltage matrix can be represented as ,in, For voltage matrix, Represents the set of real numbers, where all elements in the matrix are real numbers. T The number of rows in the matrix corresponds to the length of the sampling time series, indicating the total number of samples collected. T Data at each point in time, N The column number of the matrix corresponds to the number of battery cells in the battery, indicating the total number of cells. N Each battery cell.

[0032] In one implementation, the voltage matrix can be standardized to eliminate dimensional differences between different cells. The standardization process can be expressed as follows: ,in, For the standardized voltage matrix, σ is the mean vector calculated by column, and σ is the standard deviation vector calculated by column.

[0033] Step 102: Perform a low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix.

[0034] Low-rank approximation is a mathematical transformation method that yields a low-rank approximation matrix that preserves the most important and general variation patterns in the voltage matrix while filtering out secondary information that may be noise or subtle individual differences. The low-rank approximation matrix can characterize the collective uniformity behavior of battery cells under normal operating conditions.

[0035] In one implementation, the voltage matrix can be decomposed using singular value decomposition: ,in, Let be a left singular vector matrix, representing the eigenvectors of the time dimension; It is a singular value diagonal matrix; It is a right singular vector matrix, representing the spatial characteristic vectors between battery cells; A right singular vector matrix The transpose of .

[0036] In one implementation, a low-rank approximation matrix can be constructed by selecting the top k principal components based on the energy accumulation ratio of the singular values: Where k satisfies: , The singular value diagonal matrix is ​​the first... One singular value; Indicates the first The energy contribution of the principal component, i.e., the first principal component in the voltage matrix. The strength of each behavioral pattern contributes to the consistency structure of the battery cell group; Let be the rank of the singular value diagonal matrix, which is the number of all non-zero singular values; This represents the total energy of the system; The preset energy accumulation percentage threshold can be, for example, 0.9.

[0037] Singular value decomposition is performed on the voltage matrix. The left and right covariance matrices are decomposed to obtain the spatial structure and time-varying structure of the cell, respectively. The two are then fused by energy weighting through singular values, so that the high-dimensional battery voltage matrix can be represented as a low-rank interpretable structural combination.

[0038] Step 103: Extract the principal component projection features of each battery cell from the left or right singular vector matrix of the low-rank approximation matrix.

[0039] The voltage matrix is ​​approximated by a low-rank approximation to obtain a low-rank approximation matrix. The left singular vector matrix of the low-rank approximation matrix is ​​the first k columns of the left singular vector matrix of the voltage matrix, and the right singular vector matrix of the low-rank approximation matrix is ​​the first k columns of the right singular vector matrix of the voltage matrix, which is the first k rows of the transpose of the right singular vector matrix of the voltage matrix.

[0040] The left or right singular vector matrix of a low-rank approximation matrix can be obtained when constructing the low-rank approximation matrix, or it can be obtained by performing singular value decomposition on the low-rank approximation matrix.

[0041] Principal component projection features are used to measure the degree of participation of each cell in the main variation modes of the battery pack.

[0042] In one implementation, if the rows of the voltage matrix correspond to sampling times and the columns correspond to individual cells in the battery, then the principal component projection features of each cell are extracted from the right singular vector matrix of the low-rank approximation matrix. Conversely, if the columns of the voltage matrix correspond to sampling times and the rows correspond to individual cells in the battery, then the principal component projection features of each cell are extracted from the left singular vector matrix of the low-rank approximation matrix.

[0043] For example, if the rows of the voltage matrix correspond to the sampling time and the columns correspond to the individual cells in the battery, then from the perspective of the voltage matrix dimension, it is: transpose of the right singular vector matrix In the middle, before extraction kLine, forming dimension is The transpose of the right singular vector matrix of the low-rank approximation matrix , The multiple rows in the matrix represent the multiple principal component direction vectors of the low-rank approximation matrix. Each column represents the projection coefficients of the corresponding cell onto each principal component. Therefore, for each cell... You can calculate its column, i.e., the column number. The column is the sum of the absolute values ​​of all its elements. The principal component projection features of each battery cell are obtained. Indicates the row number. Indicates the column number.

[0044] For example, if the rows of the voltage matrix correspond to the individual cells in the battery and the columns correspond to the sampling time, then from the perspective of the voltage matrix: Extracting the front from the left singular vector k The column formation dimension is The left singular vector matrix of the low-rank approximation matrix U k , U k The multiple columns in the matrix are the multiple principal component direction vectors of the low-rank approximation matrix. U k Each row in the table represents the projection coefficients of the corresponding cell onto each principal component. Therefore, for each cell... j It can calculate its row, i.e., the row number. j The sum of the absolute values ​​of all elements yields the principal component projection characteristics for each cell.

[0045] When all cells in the battery are functioning normally, the principal component projection characteristics of each cell are as follows: The weights should be similar because when all cells are functioning normally, the collective behavior is contributed by all cells, with a uniform weight distribution. However, when a cell in the battery fails, its voltage change pattern differs from the collective behavior, leading to a significant shift in the weights of the faulty cell on certain principal components. It is significantly different from other battery cells.

[0046] In one implementation, the sum of the absolute values ​​or the sum of the squares of the corresponding row or column elements in the left or right singular vector matrices of the low-rank approximation matrix can be used as the principal component projection feature of the corresponding battery cell. The absolute value accurately determines the strength of the battery cell's participation in the corresponding principal component, while ignoring the direction of participation. The sum of squares amplifies the influence of projection coefficients with larger absolute values, making it more sensitive to the significant contribution of the battery cell to a particular principal component.

[0047] Step 104: Calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature for each cell from the residual matrix.

[0048] The residual matrix can be obtained by calculating the element-wise difference between the voltage matrix and the low-rank approximation matrix. Each element in the residual matrix represents the degree of voltage deviation of the corresponding cell at a given time. For example, the residual matrix can be represented as follows: , ,in, Represents the residual matrix. Represents the voltage matrix. Represents a low-rank approximate matrix. Indicates the length of the time series. Indicates the number of battery cells.

[0049] In one implementation, the residual feature can be a statistic extracted from the residual matrix to describe the degree to which each cell deviates from the group's behavioral pattern. The residual feature can include at least one of the following: mean residual feature, maximum residual feature, and residual standard deviation feature, wherein the mean residual feature is the average value of the cell residual time series in the matrix, characterizing the overall deviation of the cells; the maximum residual feature is the maximum value of the cell residual time series in the matrix, characterizing the instantaneous extreme anomaly of the cell; and the residual standard deviation feature is the standard deviation of the cell residual time series in the matrix, characterizing the stability of cell fluctuations.

[0050] Step 105: The residual features and principal component projection features of each cell are weighted and fused according to their respective weights to obtain the fault monitoring features of each cell.

[0051] The weights of each residual feature and principal component projection feature can be set based on engineering experience or learned from historical data through optimization algorithms or machine learning methods. The weights reflect the importance of the feature in battery fault monitoring. For each cell, each feature of the cell is multiplied by its corresponding weight coefficient to obtain a weighted value. Then, the weighted values ​​of all features of the cell are summed to obtain a total, which represents the fault monitoring feature of the cell.

[0052] Step 106: Identify faulty cells in the battery based on the fault monitoring characteristics of all cells.

[0053] When a battery pack is operating normally, the fault monitoring characteristic values ​​of most cells should be within a relatively concentrated and stable range. However, the fault monitoring characteristic values ​​of cells that are actually malfunctioning will deviate significantly from this distribution. Therefore, by analyzing the fault monitoring characteristics of all cells, faulty cells in the battery can be identified.

[0054] In one implementation, the average and standard deviation of all cell fault monitoring characteristic values ​​can be calculated, and then the value of each cell can be compared with the average. If the absolute difference between the value of a cell and the average exceeds a certain multiple of the standard deviation, such as three times, then the cell is identified as a faulty cell.

[0055] For example, for the first i For each battery cell, if its fault monitoring characteristics satisfy If so, the cell is determined to be an abnormal or faulty cell. This represents the average value of the fault monitoring characteristics of all battery cells. The standard deviation of all cell fault monitoring characteristics.

[0056] In another implementation, the fault monitoring characteristic values ​​of all cells can be sorted from smallest to largest to determine the first quartile Q1 located at 1 / 4, the third quartile Q3 located at 3 / 4, and the interquartile range IQR. Then, cells located below the first quartile Q1 minus 1.5 times the interquartile range IQR, or above the third quartile Q3 plus 1.5 times the interquartile range IQR, are identified as faulty cells.

[0057] For example, for the first i For each battery cell, if its fault monitoring characteristics satisfy or satisfy If so, the cell is determined to be an abnormal or faulty cell.

[0058] The battery fault monitoring method provided in this application obtains the voltage matrix of the battery during operation, performs low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix, which can accurately reflect the group consistency behavior of the cells in the battery under the current operating conditions and adapt to various changing operating environments. From the left singular vector matrix or the right singular vector matrix of the low-rank approximation matrix, the principal component projection features of each cell are extracted, which can reflect the projection features of each cell in the main behavior patterns of all cells in the battery. The residual matrix of the voltage matrix and the low-rank approximation matrix is ​​calculated, and at least one residual feature of each cell is extracted from the residual matrix, which can reflect the small voltage deviation of the early fault of the cell from the group consistency. The residual features and principal component projection features of each cell are weighted and fused according to their respective weights to obtain the fault monitoring features of each cell, which can reflect the true health status of the cell from multiple dimensions. Then, based on the fault monitoring features of all cells, the faulty cells in the battery can be identified, and cells in the early stage of fault in the battery can be accurately identified.

[0059] In some embodiments, in order to obtain a more accurate low-rank approximation matrix, the processing procedure of step 102 may include steps 201 to 204.

[0060] Step 201: Perform singular value decomposition on the voltage matrix to obtain the left singular vector matrix, the singular value diagonal matrix, and the transpose of the right singular vector matrix of the voltage matrix.

[0061] When the rows of the voltage matrix correspond to sampling times and the columns correspond to individual cells in the battery, each column of the left singular vector matrix corresponds to a major group behavior pattern, and each row corresponds to a sampling time. The elements in the left singular vector matrix can characterize the activity level or intensity of the corresponding behavior pattern at a specific point in time. Because the left singular vector matrix is ​​associated with the time dimension, it can reflect the shape of voltage changes over time in multiple cells within the battery.

[0062] The elements on the diagonal of a singular value diagonal matrix are called singular values, and all off-diagonal elements are zero. Singular values ​​are non-negative real numbers and are arranged in descending order on the diagonal. The magnitude of a singular value represents the energy share of its corresponding behavioral pattern in the voltage matrix.

[0063] In the transpose of the right singular vector matrix, each row corresponds to a primary group behavior pattern, and each column corresponds to a battery cell. The values ​​of the elements in the transpose of the right singular vector matrix can characterize the participation weight of each battery cell in different behavior patterns.

[0064] The process of singular value decomposition can be found in the description in the above embodiments, and will not be repeated here.

[0065] Step 202: Determine the energy percentage of the principal component corresponding to each singular value in the voltage matrix.

[0066] A principal component is an independent, identifiable pattern of change in a voltage matrix. A principal component is defined by a pair of left singular vectors, a singular value, and a right singular vector. The first principal component corresponds to the largest singular value, representing the strongest and most dominant overall trend of change in the matrix; the energy of subsequent principal components decreases sequentially.

[0067] For each principal component, its energy percentage in the voltage matrix is ​​equal to the square of the i-th singular value divided by the sum of the squares of all singular values, i.e. ,in, The singular value diagonal matrix is ​​the first... One singular value; Indicates the first The energy of each principal component; Let be the rank of the singular value diagonal matrix.

[0068] Step 203: Based on the energy proportion of each principal component in the voltage matrix, select the top principal components from the multiple principal components of the voltage matrix that satisfy the preset cumulative energy proportion condition. k One target principal component.

[0069] The preset cumulative energy percentage condition can be sorted by singular values ​​from largest to smallest, with the first one listed first. k The sum of the energy percentages of the target principal components exactly reaches or exceeds the preset cumulative energy percentage for the first time, such as 90% or 95%.

[0070] In one implementation, the energy percentage of each principal component can be accumulated sequentially in descending order of singular values. Accumulation stops when the accumulated sum first reaches or exceeds a preset accumulated energy percentage. The number of principal components accumulated at this point is recorded, denoted as . k , obtained before k The target principal components.

[0071] Step 204, the front of the left singular vector matrix of the voltage matrix. k The first column of the singular values ​​of the voltage matrix in the diagonal matrix. k The transpose of the first singular value and the right singular vector matrix of the voltage matrix. k Multiplying the rows yields a low-rank approximate matrix.

[0072] The low-rank approximation matrix is ​​used to characterize the cell group consistency behavior of the battery pack under normal operating conditions.

[0073] In one implementation, the front of the left singular vector matrix can be... k Columns, the first column of a singular value diagonal matrix k The first singular value and the transpose of the right singular vector matrix. k Multiplying the rows of the truncated matrices yields a low-rank approximation matrix of the voltage matrix. ,in, k satisfy , The singular value diagonal matrix is ​​the first... One singular value; Indicates the first The energy percentage of the principal component, i.e. the first principal component The magnitude of the contribution of the strength of each behavioral pattern to the consistency structure of the cell group; Let be the rank of the singular value diagonal matrix, which is the number of all non-zero singular values; The total energy of all battery cells; This is the preset cumulative energy percentage.

[0074] The battery fault monitoring method provided in this application performs singular value decomposition on the voltage matrix to obtain the left singular vector matrix, the singular value diagonal matrix, and the transpose of the right singular vector matrix, which can accurately reflect the consistent behavior of the cell group. It automatically determines the number of principal components to retain by calculating the energy percentage of each principal component and based on a preset threshold. k It can adapt to the data distribution of different battery packs and operating conditions, and thus the front of the left singular vector matrix of the voltage matrix. k The first column of the singular values ​​of the voltage matrix in the diagonal matrix. k The transpose of the first singular value and the right singular vector matrix of the voltage matrix. k Multiplying rows yields a low-rank approximation matrix, which enables the constructed low-rank model to accurately reflect the normal behavior pattern of the battery cell.

[0075] In some embodiments, in order to accurately extract the principal component projection features of each cell, the processing procedure of step 103 may include steps 301 to 302.

[0076] Step 301: Calculate the sum of the absolute values ​​of the row or column elements corresponding to each cell in the left or right singular vector matrix of the low-rank approximation matrix; Step 302: Determine the principal component projection features of each battery cell based on the sum of the absolute values ​​of the elements of the corresponding row or column in the left or right singular vector matrix for each battery cell.

[0077] Principal component projection features are used to characterize the projected position of an individual cell in the subspace of normal group behavior. Under normal operating conditions, cells in a battery exhibit significant group consistency, and their changing trends can be characterized by a low-dimensional dominant subspace. When an early anomaly occurs in a cell, it often manifests first as a shift in its projected position within this dominant subspace, rather than immediately producing a significant reconstruction error. Therefore, principal component projection features can be used to characterize the degree to which cells participate in the coordinated changes within the group, thus enabling sensitive characterization of early, weak anomalies.

[0078] For example, the rows of the voltage matrix correspond to the sampling time (length is...). T ), corresponding to each cell in the battery pack (the number is N Then, after performing singular value decomposition on the voltage matrix, the transpose V of the right singular vector matrix is ​​obtained. T (dimension is) N × N Extracting the previous step from the previous step) k Line, forming dimension is k × N The transpose V of the right singular vector matrix of the low-rank approximation matrix k T In V kT In this model, each row corresponds to a retained principal component (i.e., a group behavior pattern), and each column corresponds to a single cell in the battery pack. Each element's value represents the projection coefficient or weight of the cell in that column onto the principal component corresponding to that row. Therefore, for each cell, V can be... k T The corresponding column k The sum of the absolute values ​​of each element determines the principal component projection characteristic of each cell, in order to quantify the combined contribution of each cell to all major voltage variation modes.

[0079] For example, the rows of the voltage matrix correspond to the individual cells in the battery pack (the number of cells is...). N ), column corresponding to sampling time (length is T Then, after performing singular value decomposition on the voltage matrix, the left singular vector matrix U (with dimension ) is... N × N Extracting the previous step from the previous step) k Columns, forming dimensions N × k The left singular vector matrix U of the low-rank approximation matrix k In this matrix, each row corresponds to a single cell in the battery pack, and each column corresponds to a retained principal component (i.e., a group behavior pattern). Each element's value represents the projection coefficient or weight of the cell in that row onto the principal component in that column. Therefore, for each cell, U... k The corresponding row in the middle k The sum of the absolute values ​​of each element determines the principal component projection characteristic of each cell, in order to quantify the combined contribution of each cell to all major voltage variation modes.

[0080] In one implementation, L1 can be the sum of the absolute values ​​of the elements of the corresponding row or column in the transpose of the left or right singular vector matrix of the low-rank approximation matrix for each cell. i The principal component projection features of each battery cell are determined. Since the elements in the transpose of the right singular vector matrix can be either positive or negative, taking the absolute value can accurately determine the participation intensity of the battery cell on the corresponding principal component, while ignoring the direction of participation.

[0081] The sum of the absolute values ​​L1 of the corresponding row or column elements in the transpose of the left or right singular vector matrix of a low-rank approximation matrix. iThe sum of the absolute values ​​of the elements represents the contribution strength of the corresponding cell to all retained major collective behavior patterns of the battery pack. The larger the sum of the absolute values ​​of the elements, the more consistent the behavior of the corresponding cell is with the major collective patterns of the battery pack, and the higher its health level; conversely, the smaller the sum of the absolute values ​​of the elements, the lower the participation of the corresponding cell in the major patterns, and the more likely it has become abnormal. Therefore, the sum of the absolute values ​​of the elements of the corresponding column in the left singular vector matrix or the transpose of the right singular vector matrix of the low-rank approximation matrix for each cell is determined as the principal component projection feature of each cell.

[0082] In one implementation, the sum of the absolute values ​​of each cell can be divided by the median or maximum of the sum of the absolute values ​​of all cells to obtain a relative value, and the relative value can be used to determine the principal component projection feature of each cell.

[0083] In one implementation, L1 can be the sum of the absolute values ​​of the cell in the current time window and the previous time window. i The difference is determined as the principal component projection feature of the battery cell.

[0084] The battery fault monitoring method provided in this application calculates the sum of the absolute values ​​of the elements in each row or column of the transpose of the left singular vector matrix or the right singular vector matrix of the low-rank approximation matrix. It has low computational complexity and is suitable for real-time online fault monitoring of batteries. Based on the sum of the absolute values ​​of the elements in the corresponding row or column of the transpose of the left singular vector matrix or the right singular vector matrix of the low-rank approximation matrix for each cell, the principal component projection characteristics of each cell are determined, which can quantify the participation intensity of the cell in all major group behavior patterns.

[0085] In some embodiments, in order to accurately extract the residual features of the battery cell, the processing procedure of step 104 may include steps 401 to 402.

[0086] Step 401: Calculate the mean, maximum value, and standard deviation of each column element in the residual matrix.

[0087] In the residual matrix, each column corresponds to the residual data of a single battery cell at all time points, including... T A number, T This represents the total number of sampling time points.

[0088] The average value of each column of elements can reflect the overall deviation of the battery cell, that is, the average degree to which the battery cell deviates from the group's consistent behavior over the entire time period.

[0089] The maximum value of each element in the column can reflect the instantaneous extreme anomaly of the battery cell, that is, the maximum instantaneous deviation of the battery cell at a certain moment, which is used to capture sudden or extreme anomalies.

[0090] The standard deviation of each element reflects the fluctuation stability of the battery cell, that is, the degree of fluctuation of the cell's residuals over time, and the stability of its deviation behavior. The larger the standard deviation, the more unstable the voltage behavior of the battery cell.

[0091] Step 402: Based on the mean, maximum and standard deviation of the elements in the corresponding column of each cell in the residual matrix, determine the mean residual feature, maximum residual feature and residual standard deviation feature corresponding to each cell.

[0092] In one implementation, the average value of the elements in the corresponding column of the residual matrix for each cell can be determined as the average residual characteristic of the cell, the maximum value of the elements in the corresponding column of the residual matrix for each cell can be determined as the maximum residual characteristic of the cell, and the standard deviation of the elements in the corresponding column of the residual matrix for each cell can be determined as the residual standard deviation characteristic of the cell.

[0093] In another implementation, in order to eliminate scale differences between cells or between different features, the mean, maximum and standard deviation of the elements in the corresponding column of the residual matrix of all cells can be normalized. The mean, maximum and standard deviation of all cells are linearly mapped to the interval [0,1], and the normalized mean, maximum and standard deviation are determined as the mean residual feature, maximum residual feature and residual standard deviation feature corresponding to the cell, respectively.

[0094] When the residual characteristics of the battery cell include the average residual characteristics, the maximum residual characteristics, and the residual standard deviation characteristics, step 105 can be expressed as: , .

[0095] in, i Indicates the index number of the battery cell. j Indicates the index number of the principal component. These are the weighting coefficients for the principal component projection feature, the average residual feature, the maximum residual feature, and the residual standard deviation feature, respectively. , , , and The first i Fault monitoring characteristics, average residual characteristics, maximum residual characteristics, residual standard deviation characteristics, and principal component projection characteristics of individual battery cells.

[0096] Principal component projection characteristics reflect the cell's response capability in the dominant change mode of the population. This dominant mode corresponds to the synergistic change pattern of the battery system under normal operating conditions, formed by the combined effects of electrochemical reaction rate, internal resistance consistency, and thermal coupling. When an early anomaly occurs in the cell, the cell's ability to participate in the synergistic change of the population decreases, manifested as a shift in its projection in the principal component space. Residual characteristics reflect local anomalous components in the cell's behavior that cannot be explained by the dominant change mode, such as enhanced local polarization, abrupt changes in internal resistance, or abnormal self-discharge. Among these, the average residual is used to characterize the degree of chronic degradation; the maximum residual is used to characterize local sudden anomalies; and the residual dispersion is used to reflect the degree of disruption of electrochemical consistency.

[0097] While residual features alone can identify significantly deviating samples, they suffer from lag in detecting early-stage anomalous cells that are still within low-rank subspaces but have already undergone structural shifts. Principal component projection features alone fail to effectively characterize localized abrupt anomalies, resulting in insufficient sensitivity to sudden faults. Therefore, a weighted fusion of the residual features and principal component projection features of each cell, according to their respective weights, yields fault monitoring features for each cell. These features reflect the cell's evolution from a homogeneous, cooperative reactive system to a non-uniform, unstable system, thus balancing group consistency shifts and localized structural anomalies, as well as gradual degradation and sudden anomalies. This improves sensitivity to early, weak anomalies and the comprehensiveness of anomaly detection, enhances the ability to locate anomalous cells, and enables more precise differentiation of anomaly sources.

[0098] The battery fault monitoring method provided in this application embodiment uses an average value to reflect the overall deviation level of the battery cell and capture persistent consistency deviations; a maximum value to reflect the instantaneous extreme deviation of the battery cell and is sensitive to sudden faults; and a standard deviation to reflect the fluctuation stability of the battery cell and identify abnormal states with voltage fluctuations. Based on the average value, maximum value, and standard deviation of the elements in the corresponding column of the residual matrix for each battery cell, the method determines the average residual characteristic, maximum residual characteristic, and residual standard deviation characteristic for each battery cell. This allows for a description of the battery cell's health status from different dimensions, enhancing the ability to identify early, subtle anomalies through information complementarity and more accurately identifying early battery cell faults. Furthermore, the calculation complexity of the average value, maximum value, and standard deviation is low, requiring minimal computational resources, and enabling real-time monitoring of battery cell faults.

[0099] In some embodiments, in order to accurately identify the faulty battery cell, the processing procedure of step 106 includes steps 501 to 502.

[0100] Step 501: Perform outlier detection calculations on the fault monitoring characteristics of all cells to obtain the outlier value for each cell.

[0101] In one implementation, the statistical characteristics of the fault monitoring features of all cells can be determined, such as the central tendency and dispersion of the data. Then, the relative deviation of the fault monitoring features of each cell, i.e., the outlier, can be evaluated based on the statistical characteristics of the fault monitoring features of all cells.

[0102] For example, all cell fault monitoring characteristics can be calculated. arithmetic mean and standard deviation . No. i Outliers in a single battery cell can be It can also be .

[0103] Step 502: Identify the cells whose outliers meet the preset fault conditions as faulty cells.

[0104] After obtaining the outlier value for each cell, for each cell, it can be determined whether the outlier value meets a preset fault condition. If it does, the cell is identified as a faulty cell; otherwise, it is identified as a normal cell. For example, the preset fault condition could be that a cell with an outlier value greater than a preset outlier threshold is a faulty cell. Therefore, for each cell, if the outlier value reaches or exceeds the preset outlier threshold, the cell is identified as a faulty cell.

[0105] Continuing with the previous example, the preset fault condition could be an outlier value in the battery cell. .

[0106] The battery fault monitoring method provided in this application performs outlier detection calculations on the fault monitoring characteristics of all battery cells to obtain the outlier value of each battery cell. The battery cells whose outlier values ​​meet the preset fault conditions are identified as faulty battery cells. It can determine anomalies based on the distribution of fault monitoring characteristics of all battery cells in the battery, adapt to the normal fluctuation of characteristic values ​​under different batteries, different aging levels, and different operating conditions, avoid false alarms or missed alarms caused by fixed thresholds, and improve the accuracy of fault monitoring.

[0107] In some embodiments, in order to accurately identify the faulty cell, the process of step 501 may include steps 601 to 604.

[0108] Step 601: Determine the median of the fault monitoring characteristics for all cells.

[0109] The median is the middle value when a set of values ​​is arranged in ascending order. If the number of values ​​is odd, the median is the middle number; if the number of values ​​is even, the median is the average of the two middle values. The median is a statistic that describes the central tendency of a set of data. Compared to the arithmetic mean, the median is not sensitive to extreme values ​​(outliers) and is robust.

[0110] Step 602: Calculate the difference between the fault monitoring characteristics and the median of each cell to obtain the deviation sequence.

[0111] For each cell, the median of all cell fault monitoring characteristics is subtracted from the cell's fault monitoring characteristics to obtain the difference between each cell's fault monitoring characteristic and the median. This difference reflects the direction and magnitude of the deviation of the cell's fault monitoring characteristics from the typical values ​​of the entire cell group. The differences between each cell's fault monitoring characteristics and the median are then sorted according to the cell's index number, forming a sequence of length [length missing]. N The numerical array is used to obtain the deviation sequence, which reflects the deviation of the fault monitoring characteristics of all cells from the median.

[0112] Step 603: Determine the median of the deviation sequence as the absolute deviation of the median.

[0113] The absolute value of each value in the deviation sequence is taken to obtain a set of non-negative absolute values. Then, the median of this set of non-negative absolute values ​​is determined; this value is the median absolute deviation. The median absolute deviation quantifies the typical deviation of all cell fault monitoring characteristics relative to the median. Compared to the standard deviation, the median absolute deviation has a stronger ability to resist outlier interference.

[0114] Step 604: Calculate the outlier value for each cell based on the median absolute deviation and the preset scaling factor.

[0115] The preset scaling factor is a pre-defined constant used to adjust the absolute deviation of the median to a scale comparable to the standard deviation. Assuming the data follows a normal distribution, a preset scaling factor of 1.4826 can be used to ensure that the absolute deviation of the median consistently estimates the standard deviation.

[0116] In one implementation, the calculation process for the outlier value of each cell can be represented as follows: , ,in, For the first Outliers in individual cells; This represents the median of the fault monitoring characteristics for all battery cells. This represents the absolute deviation of the median.

[0117] When one or more faulty cells are present in a battery, the fault monitoring characteristics of the faulty cells differ significantly from those of normal cells. The mean and standard deviation of the fault monitoring characteristics of all cells are severely skewed by the faulty cells, causing normal cells to be misclassified as abnormal. In contrast, the median and median absolute deviation are minimally affected by extreme values ​​and are more robust to outliers. Even with multiple outliers in the data, they can accurately estimate the central location and dispersion of the normal cell group, thereby identifying the faulty cells and avoiding missed and false alarms.

[0118] The battery fault monitoring method provided in this application uses a combination of median absolute deviation (MAD) and Z-score to calculate the outlier value of the battery cell. MAD constructs a robust baseline of normal behavior, which can resist the interference of abnormal samples on the statistical center. Z-score standardizes and amplifies the degree of deviation, which can enhance the detection capability of progressive abnormal changes. Thus, it can improve the identification capability of early weak anomalies while maintaining stability, and is suitable for anomaly detection in scenarios of uniform degradation of battery cell groups.

[0119] In some embodiments, the preset fault conditions include at least one preset outlier range and its corresponding fault level, and the processing procedure of step 106 includes step 701.

[0120] Step 701: Determine the fault level of the battery cell based on the preset outlier range into which the outlier falls.

[0121] In one implementation, the outlier value of the battery cell can be compared with the boundaries of various preset outlier value ranges to find the preset outlier value range into which the outlier value falls. For example, when the outlier value of the battery cell... If so, the battery cell is determined to be abnormal or faulty. The outlier threshold can be divided into three levels. A higher outlier threshold value, while still meeting the judgment criteria, indicates a more severe anomaly or fault. Therefore, based on the outlier value of the battery cell... Quantitatively assess the severity of cell abnormalities or faults, such as: If so, it is a Level 1 fault; If so, it is a level 2 fault; If so, it is a level 3 fault, in which case, .

[0122] The battery fault monitoring method provided in this application embodiment can classify the degree of abnormality or fault of the battery cell into multiple levels by using multiple preset outlier ranges, quantify the severity of the battery cell fault, and enable the fault warning to provide richer information. It is suitable for the detection of abnormal or faulty battery cells in battery packs.

[0123] Figure 2This is a flowchart illustrating a battery fault monitoring method provided in another embodiment of this application, as shown below. Figure 2 As shown, in another embodiment of this application, the battery fault monitoring method includes steps 801 to 808.

[0124] Step 801: Obtain the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery.

[0125] Step 802: Perform a low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix.

[0126] Step 803: Calculate the sum of the absolute values ​​of the elements of the row or column corresponding to each cell in the left singular vector matrix or the right singular vector matrix.

[0127] Step 804: Determine the principal component projection features of each battery cell based on the sum of the absolute values ​​of the elements in the corresponding row or column of the left singular vector matrix or the right singular vector matrix.

[0128] Step 805: Calculate the mean, maximum, and standard deviation of the row or column elements corresponding to each cell in the residual matrix.

[0129] Step 806: Based on the mean, maximum and standard deviation of the elements in the corresponding row or column of each cell in the residual matrix, determine the mean residual feature, maximum residual feature and residual standard deviation feature corresponding to each cell.

[0130] Step 807: The residual features and principal component projection features of each cell are weighted and fused according to their respective weights to obtain the fault monitoring features of each cell.

[0131] Step 808: Identify faulty cells in the battery based on the fault monitoring characteristics of all cells.

[0132] The specific principles and implementation processes of steps 801 to 808 can be found in the descriptions in the above embodiments, and will not be repeated here.

[0133] Figure 3 This is a schematic diagram of the structure of a battery fault monitoring device provided in another embodiment of this application. Figure 3 As shown, in another embodiment of this application, the battery fault monitoring device 30 includes: The acquisition module 31 is used to acquire the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. Processing module 32 is used to perform low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix; Extraction module 33 is used to extract the principal component projection features of each cell from the left singular vector matrix or the right singular vector matrix of the low-rank approximation matrix. Calculation module 34 is used to calculate the residual matrix of the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature of each cell from the residual matrix. The fusion module 35 is used to perform weighted fusion of the residual features and principal component projection features of each cell according to their respective weights to obtain the fault monitoring features of each cell. The identification module 36 is used to identify faulty cells in the battery based on the fault monitoring characteristics of all cells.

[0134] In some embodiments, when processing module 32 performs low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix, it specifically performs: singular value decomposition on the voltage matrix to obtain the transpose of the left singular vector matrix, the singular value diagonal matrix, and the right singular vector matrix of the voltage matrix; determines the energy proportion of the principal component corresponding to each singular value in the singular value diagonal matrix in the voltage matrix; and selects the top principal components of the voltage matrix that exactly satisfy the preset cumulative energy proportion condition from the multiple principal components of the voltage matrix according to the energy proportion of each principal component in the voltage matrix. k One target principal component; the left singular vector matrix of the voltage matrix is ​​the first... k The first column of the singular values ​​of the voltage matrix in the diagonal matrix. k The transpose of the first singular value and the right singular vector matrix of the voltage matrix. k Multiplying the rows yields a low-rank approximate matrix.

[0135] In some embodiments, when the extraction module 33 extracts the principal component projection features of each battery cell from the left singular vector matrix or the right singular vector matrix, it is specifically used to: calculate the sum of the absolute values ​​of the elements of the corresponding row or column of each battery cell in the left singular vector matrix or the right singular vector matrix; and determine the principal component projection features of each battery cell based on the sum of the absolute values ​​of the elements of the corresponding row or column of each battery cell in the left singular vector matrix or the right singular vector matrix.

[0136] In some embodiments, when the calculation module 34 extracts at least one residual feature of each cell from the residual matrix, it is specifically used to: calculate the mean, maximum and standard deviation of each column element in the residual matrix; and determine the mean residual feature, maximum residual feature and residual standard deviation feature corresponding to each cell based on the mean, maximum and standard deviation of the elements in the corresponding column of the residual matrix for each cell.

[0137] In some embodiments, when the identification module 36 is used to identify faulty cells in the battery based on the fault monitoring characteristics of all cells, it is specifically used to: perform outlier detection calculation on the fault monitoring characteristics of all cells to obtain the outlier value of each cell; and determine the cells whose outlier values ​​meet the preset fault conditions as faulty cells.

[0138] In some embodiments, when the identification module 36 performs outlier detection calculations on the fault monitoring features of all battery cells to obtain the outlier value of each battery cell, it is further configured to: determine the median of the fault monitoring features of all battery cells; calculate the difference between the fault monitoring features of each battery cell and the median to obtain a deviation sequence; determine the median of the deviation sequence as the absolute deviation of the median; and calculate the outlier value of each battery cell based on the absolute deviation of the median and a preset scaling factor.

[0139] In some embodiments, the preset fault conditions include at least one preset outlier range and its corresponding fault level. When the identification module 36 is used to determine the battery cell whose outlier meets the preset fault conditions as a faulty battery cell, it is further used to: determine the fault level corresponding to the battery cell according to the preset outlier range into which the outlier falls.

[0140] The battery fault monitoring device provided in this application embodiment can execute the battery fault monitoring method provided in any of the above embodiments. The specific implementation method and principle are similar, and will not be described in detail here.

[0141] This application also provides a battery management system, including: The sampling circuit is used to acquire the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. The control circuit is used to perform low-rank approximation on the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix; extract the principal component projection features of each cell from the left or right singular vector matrix of the low-rank approximation matrix; calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature of each cell from the residual matrix; perform weighted fusion of the residual features and principal component projection features of each cell according to their respective weights to obtain the fault monitoring features of each cell; and identify the faulty cells in the battery based on the fault monitoring features of all cells.

[0142] In one implementation, the control circuit is further configured to: perform singular value decomposition on the voltage matrix to obtain the transpose of the left singular vector matrix, the singular value diagonal matrix, and the right singular vector matrix of the voltage matrix; determine the energy proportion of the principal component corresponding to each singular value in the singular value diagonal matrix in the voltage matrix; and select the principal components that satisfy the preset cumulative energy proportion condition from the multiple principal components of the voltage matrix according to the energy proportion of each principal component in the voltage matrix. kOne target principal component; the left singular vector matrix of the voltage matrix is ​​the first... k The first column of the singular values ​​of the voltage matrix in the diagonal matrix. k The transpose of the first singular value and the right singular vector matrix of the voltage matrix. k Multiplying the rows yields a low-rank approximate matrix.

[0143] In one implementation, the control circuit is further configured to: calculate the sum of the absolute values ​​of the elements of the corresponding row or column of each cell in the left singular vector matrix or the right singular vector matrix; and determine the principal component projection features of each cell based on the sum of the absolute values ​​of the elements of the corresponding row or column of each cell in the left singular vector matrix or the right singular vector matrix.

[0144] In one implementation, the control circuit is further used to: calculate the mean, maximum and standard deviation of each column element in the residual matrix; and determine the mean residual characteristic, maximum residual characteristic and residual standard deviation characteristic corresponding to each cell based on the mean, maximum and standard deviation of the elements in the corresponding column of the residual matrix for each cell.

[0145] In one implementation, the control circuit is further configured to: perform outlier detection calculations on the fault monitoring characteristics of all cells to obtain the outlier value of each cell; and identify cells whose outlier values ​​meet preset fault conditions as faulty cells.

[0146] In one implementation, the control circuit is further configured to: determine the median of the fault monitoring characteristics of all cells; calculate the difference between the fault monitoring characteristics of each cell and the median to obtain a deviation sequence; determine the median of the deviation sequence as the absolute deviation of the median; and calculate the outlier value of each cell based on the absolute deviation of the median and a preset scaling factor.

[0147] In one implementation, the preset fault conditions include at least one preset outlier range and its corresponding fault level. The control circuit is further configured to: determine the fault level corresponding to the cell based on the preset outlier range into which the outlier falls.

[0148] This application also provides a battery device, which includes multiple battery cells and any of the above-described battery management systems.

[0149] This application also provides an electrical device, which includes any of the battery devices described above.

[0150] Figure 4 This is a schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application. For example... Figure 4 As shown, the electronic device may include a processor 41 and a memory 42 storing computer program instructions.

[0151] Specifically, the processor 41 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0152] Memory 42 may include mass storage for data or instructions. For example, and not limitingly, memory 42 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 42 may include removable or non-removable (or fixed) media. Where appropriate, memory 42 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 42 is non-volatile solid-state memory.

[0153] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Thus, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0154] The processor 41 reads and executes computer program instructions stored in the memory 42 to implement any of the battery fault monitoring methods in the above embodiments.

[0155] In one example, the electronic device may also include a communication interface 43 and a bus 44. The processor 41, memory 42, and communication interface 43 are connected via the bus 44 and communicate with each other.

[0156] Communication interface 43 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0157] Bus 44 includes hardware, software, or both, that couples components of an electronic device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 44 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, this application contemplates any suitable bus or interconnect.

[0158] Furthermore, in conjunction with the battery fault monitoring methods in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the battery fault monitoring methods in the above embodiments.

[0159] This application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the battery fault monitoring methods described in the above embodiments.

[0160] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0161] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. The programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0162] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0163] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowchart illustrations, and combinations of blocks in the block diagrams and / or flowchart illustrations, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0164] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A battery fault monitoring method, characterized in that, include: Obtain the voltage matrix of the battery during operation, the voltage matrix being used to characterize the correspondence between sampling time and the voltage of multiple cells in the battery; Singular value decomposition is performed on the voltage matrix to obtain the left singular vector matrix, the singular value diagonal matrix, and the transpose of the right singular vector matrix of the voltage matrix; Determine the energy percentage of the principal component corresponding to each singular value in the voltage matrix; Based on the energy proportion of each principal component in the voltage matrix, select the top k target principal components that satisfy the preset cumulative energy proportion condition from multiple principal components of the voltage matrix; Multiply the first k columns of the left singular vector matrix of the voltage matrix, the first k singular values ​​in the singular value diagonal matrix of the voltage matrix, and the first k rows of the transpose of the right singular vector matrix of the voltage matrix to obtain the low-rank approximate matrix of the voltage matrix. Calculate the sum of the absolute values ​​of the elements in the row or column corresponding to each cell in the left singular vector matrix or the right singular vector matrix; The principal component projection features of each battery cell are determined based on the sum of the absolute values ​​of the elements of the corresponding row or column in the left or right singular vector matrix. Calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature for each cell from the residual matrix; The residual features and principal component projection features of each cell are weighted and fused according to their respective weights to obtain the fault monitoring features of each cell. Based on the fault monitoring characteristics of all cells, faulty cells in the battery are identified.

2. The battery fault monitoring method according to claim 1, characterized in that, Extracting at least one residual feature for each cell from the residual matrix includes: Calculate the mean, maximum, and standard deviation of the row or column elements corresponding to each cell in the residual matrix; Based on the mean, maximum, and standard deviation of the elements in the corresponding row or column of each cell in the residual matrix, the mean residual characteristic, maximum residual characteristic, and residual standard deviation characteristic of each cell are determined.

3. The battery fault monitoring method according to claim 1, characterized in that, The step of identifying faulty cells in the battery based on the fault monitoring characteristics of all cells includes: Outlier detection calculations are performed on the fault monitoring characteristics of all cells to obtain the outlier value for each cell; Cells whose outliers meet preset fault conditions are identified as faulty cells.

4. The battery fault monitoring method according to claim 3, characterized in that, The outlier detection calculation for the fault monitoring characteristics of all battery cells to obtain the outlier value for each battery cell includes: Determine the median of the fault monitoring characteristics for all battery cells; The difference between the fault monitoring characteristics of each cell and the median is calculated to obtain the deviation sequence; Determine the median of the deviation sequence as the absolute median deviation; The outlier value for each cell is calculated based on the median absolute deviation and the preset scaling factor.

5. The battery fault monitoring method according to claim 4, characterized in that, The preset fault conditions include at least one preset outlier range and its corresponding fault level. Determining a cell whose outlier meets the preset fault conditions as a faulty cell includes: The fault level of the battery cell is determined based on the preset outlier range into which the outlier falls.

6. A battery management system, characterized in that, include: A sampling circuit is used to acquire the voltage matrix of the battery during operation. The voltage matrix is ​​used to characterize the correspondence between the sampling time and the voltage of multiple cells in the battery. A control circuit is used to perform singular value decomposition on the voltage matrix to obtain the left singular vector matrix, the singular value diagonal matrix, and the transpose of the right singular vector matrix of the voltage matrix. Determine the energy percentage of the principal component corresponding to each singular value in the voltage matrix; Based on the energy proportion of each principal component in the voltage matrix, select the top k target principal components that satisfy the preset cumulative energy proportion condition from multiple principal components of the voltage matrix; multiply the first k columns of the left singular vector matrix of the voltage matrix, the first k singular values ​​in the singular value diagonal matrix of the voltage matrix, and the first k rows of the transpose of the right singular vector matrix of the voltage matrix to obtain a low-rank approximation matrix of the voltage matrix; calculate the sum of the absolute values ​​of the elements of the corresponding row or column of each cell in the left or right singular vector matrix; based on the sum of the absolute values ​​of the elements of the corresponding row or column of each cell in the left or right singular vector matrix, determine the principal component projection characteristics of each cell; Calculate the residual matrix between the voltage matrix and the low-rank approximation matrix, and extract at least one residual feature for each cell from the residual matrix; weight and fuse the residual feature of each cell with the principal component projection feature according to their respective weights to obtain the fault monitoring feature of each cell; identify the faulty cell in the battery based on the fault monitoring features of all cells.

7. A battery device, characterized in that, It includes multiple battery cells and a battery management system as described in claim 6.

8. An electrical appliance, characterized in that, Includes the battery device as described in claim 7.

Images