Battery cell internal short circuit anomaly detection method, device, and storage medium

Abstract

The application relates to the battery technical field and discloses a battery cell internal short circuit abnormality detection method, equipment and a storage medium. The method is used for detecting a plurality of battery cells and comprises the following steps: collecting the voltage, current and temperature of each battery cell at N collection time points, and determining a first matrix corresponding to each battery cell based on the collected data; performing normalization processing on the first matrix corresponding to each battery cell to obtain a second matrix corresponding to each battery cell; determining a third matrix according to the first matrices corresponding to all the battery cells; performing principal component analysis on the third matrix to determine a feature matrix used for representing the state evolution of the battery cells; determining the feature trajectory of the i-th battery cell according to the feature matrix and the second matrix corresponding to the i-th battery cell; determining the trajectory evolution speed and the trajectory evolution direction of the battery cell feature trajectory corresponding to each battery cell; and determining whether internal short circuit abnormality occurs in each battery cell according to the trajectory evolution speed and the trajectory evolution direction. The application can accurately identify whether internal short circuit abnormality occurs in the battery cells.

Description

Technical Field

[0001] This application relates to the field of battery technology, specifically to a method, device, and storage medium for detecting short circuit anomalies within a battery cell. Background Technology

[0002] An internal short circuit within a battery cell refers to the direct contact between the positive and negative electrodes due to some reason, forming a low-resistance current path. This can lead to battery overheating, performance degradation, and even fire or explosion. With the continuous increase in the capacity and energy density of energy storage systems, the instability of the internal structure of battery cells within these systems has intensified, significantly increasing the risk of internal short circuits. In the early stages of an internal short circuit (i.e., when a minor internal short circuit occurs), the cell's voltage or temperature does not change significantly, making it difficult to effectively identify using conventional voltage and temperature monitoring methods. However, at this point, the cell already exhibits latent characteristics such as accelerated self-discharge rate, slow internal resistance drift, and abnormal temperature rise characteristics. If a micro-internal short circuit is not detected in time, it is highly likely to evolve into thermal runaway, posing a serious threat to the overall safety of the energy storage system. Summary of the Invention

[0003] In view of the above problems, this application provides a method, device and storage medium for detecting internal short circuit anomalies in battery cells, which solves the problem in the prior art that it is impossible to quickly and accurately identify whether an internal short circuit anomaly has occurred in a battery cell.

[0004] According to one aspect of the embodiments of this application, a method for detecting short-circuit anomalies within a battery cell is provided for detecting multiple battery cells in an energy storage system. The method includes: acquiring the voltage, current, and temperature of each battery cell at N acquisition times, and determining an N×5 first matrix corresponding to each battery cell based on the acquired voltage, current, and temperature, wherein each row of the first matrix corresponding to the i-th battery cell is... , , and The data are the voltage, current, and temperature of the i-th cell collected at time t, with different rows of data corresponding to different collection times t. N and i are both positive integers, and N > 2. The first matrix corresponding to each cell is normalized to obtain the second matrix corresponding to each cell. An N×5 third matrix is ​​determined based on the first matrices corresponding to all cells. ,in, The j-th row of the third matrix , Each data point in the third matrix is ​​determined based on the data in the j-th row of the entire first matrix, where j is a positive integer and j≤N; Principal component analysis was performed to determine a 5×2 characteristic matrix used to characterize the cell state evolution. ; through formula Map the second matrix of each battery cell to the feature matrix. This yields the characteristic trajectory of each battery cell, where... The cell feature trajectory corresponding to the i-th cell The second matrix is ​​given for the i-th cell; the trajectory evolution speed and trajectory evolution direction of the characteristic trajectory of each cell are determined respectively; based on the trajectory evolution speed and trajectory evolution direction of the characteristic trajectory of each cell, it is determined whether an internal short circuit anomaly occurs in each cell.

[0005] In an alternative approach, the third matrix The j-th row It is obtained by taking the median of the elements at the corresponding positions in the j-th row vector of the first matrix for all battery cells.

[0006] In one alternative approach, the third matrix Principal component analysis was performed to determine a 5×2 characteristic matrix used to characterize the cell state evolution. This includes: the third matrix Perform mean-removal processing to obtain the mean-removed matrix. ; through formula Determine the covariance matrix Determine the covariance matrix. First eigenvalue Second eigenvalue Determine the first feature value The corresponding first characteristic matrix and the second eigenvalue The corresponding second characteristic matrix ; the first feature matrix and the second feature matrix The constructed matrix is ​​determined as the characteristic matrix. ,in, .

[0007] In one optional approach, the trajectory evolution rate of the characteristic trajectory of each battery cell is determined, including: for the i-th battery cell, using the formula... Determine the trajectory evolution rate of the characteristic trajectory of the i-th cell. ,in, It is the difference vector formed by two adjacent points in the characteristic trajectory of the i-th cell.

[0008] In one optional approach, the trajectory evolution direction of the characteristic trajectory of each battery cell is determined, including: using a formula Determine the trajectory evolution direction of the characteristic trajectory of the i-th cell. ,in, and These are the difference vectors formed by two adjacent points in the characteristic trajectory of the i-th cell.

[0009] In one optional approach, determining whether each battery cell has experienced an internal short-circuit anomaly based on the trajectory evolution rate and trajectory evolution direction of the characteristic trajectory corresponding to each battery cell includes: using a formula Determine the characteristic trajectory of the reference cell For the i-th cell, determine the cell feature trajectory corresponding to the i-th cell. With the reference cell characteristic trajectory DTW distance between According to the DTW distance corresponding to the i-th cell Determine the similarity between the feature trajectory of the i-th cell and the feature trajectory of the reference cell. Based on the similarity corresponding to the i-th cell The trajectory evolution speed and direction of the characteristic trajectory of the battery cell are used to determine whether the i-th battery cell has an internal short circuit anomaly.

[0010] In one alternative approach, the step is based on the DTW distance corresponding to the i-th cell. Determine the similarity between the feature trajectory of the i-th cell and the feature trajectory of the reference cell. , including: through formula Determine the similarity corresponding to the i-th cell. The similarity based on the i-th cell The trajectory evolution speed and direction of the cell characteristic trajectory are used to determine whether the i-th cell has an internal short circuit anomaly, including: for the i-th cell, using the formula... Determine the rate of change of comprehensive characteristics corresponding to the i-th cell. ,in, Preset parameters; through formula Determine the overall anomaly degree corresponding to the i-th cell. ,in, and The weights are preset; if the overall anomaly degree corresponding to the i-th cell is... If the value exceeds a preset threshold, it is determined that the i-th cell has experienced an internal short circuit anomaly.

[0011] In one alternative approach, if the overall anomaly degree corresponding to the i-th cell... If the overall anomaly is greater than a preset threshold, an internal short circuit anomaly is determined for the i-th cell, including: if the overall anomaly degree corresponding to the i-th cell is greater than a preset threshold. If the overall anomaly level of the i-th cell exceeds the first preset threshold, the level of the internal short circuit anomaly is determined to be Level 1; if the overall anomaly level of the i-th cell is... If the value is greater than the second preset threshold, the level of the internal short circuit anomaly in the i-th cell is determined to be the second level, wherein the second preset threshold is greater than the first preset threshold, and the second level is higher than the first level.

[0012] According to another aspect of the embodiments of this application, a short-circuit anomaly detection device for battery cells is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the short-circuit anomaly detection method for battery cells as described above.

[0013] According to another aspect of the embodiments of this application, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the cell short-circuit anomaly detection method as described above.

[0014] In this embodiment, a first matrix is ​​determined based on the voltage, temperature, and current of the battery cells collected at multiple times to extract multi-dimensional features that change over time. Each first matrix is ​​then normalized and time-aligned to form a time-varying feature vector sequence, ensuring comparability between different battery cells in the time domain and providing a unified scale for determining whether a battery cell has experienced an internal short-circuit anomaly based on its corresponding feature trajectory. Next, a third matrix is ​​determined based on the first matrices of all battery cells, serving as the sample matrix for normal battery cells. To determine the difference between the second matrix corresponding to each battery cell and this sample matrix, and thus whether an internal short-circuit anomaly has occurred in each battery cell, this application performs principal component analysis on the sample matrix corresponding to normal battery cells to obtain the principal component basis matrix (i.e., feature matrix P). The high-dimensional time-varying features (i.e., the second matrix) corresponding to each battery cell are projected into a low-dimensional space to form a two-dimensional trajectory (i.e., the battery cell feature trajectory), thereby achieving information compression and noise suppression, making the changes in the battery cell feature trajectory better reflect the potential abnormal trends of the battery cell. Then, based on the motion characteristics of the cell's characteristic trajectory in low-dimensional space, the local trajectory evolution velocity and trajectory evolution direction are calculated. This allows us to reveal the dynamic changes such as deflection and fluctuations of the cell's characteristic trajectory through its trajectory evolution velocity and direction. Furthermore, subsequent analysis based on the trajectory evolution velocity of the cell's characteristic trajectory... When determining whether a battery cell has an internal short circuit, the method of determining whether the battery cell has an internal short circuit based on the trajectory evolution direction can improve the accuracy of the judgment results compared to the method of determining whether the battery cell has an internal short circuit based on a single parameter (such as the battery cell's voltage, current, or temperature).

[0015] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0016] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0017] Figure 1 A flowchart illustrating the cell short-circuit anomaly detection method provided in an embodiment of this application is shown.

[0018] Figure 2 The illustration shows schematic diagrams of the characteristic trajectories of a reference cell, a normal cell, a cell with a minor internal short circuit, and a cell with a severe internal short circuit, respectively, provided in the embodiments of this application.

[0019] Figure 3 A schematic diagram of the structure of the cell short-circuit anomaly detection device provided in an embodiment of this application is shown. Detailed Implementation

[0020] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be implemented in various forms and should not be limited to the embodiments set forth herein.

[0021] To identify whether a battery cell has experienced an internal short circuit, deep learning models (such as LSTM and CNN) can be used to analyze the cell's data to determine if an internal short circuit has occurred. Specifically, data from the battery cell can be collected and labeled to obtain training data. This training data is then used to train a deep learning model. When it's necessary to identify whether a specific battery cell has experienced an internal short circuit, the data from that cell is input into the trained deep learning model, and the model's output will provide the identification result. However, obtaining training data requires labeling a large amount of raw data, which is time-consuming. Furthermore, because deep learning models have a large number of parameters and complex structures, they place high demands on the computing power and hardware resources of the devices deploying them.

[0022] In addition, big data statistical models can be used to analyze long-term operating data of battery cells. By extracting statistical distribution characteristics or constructing clustering and regression relationships, abnormal samples that deviate from the normal pattern can be effectively identified based on the degree of deviation of the data from the distribution center or the distance between clusters, thereby realizing the identification of micro-internal short circuits in battery cells. However, this method has high requirements for data consistency, is easily affected by the environment and equipment aging, and cannot detect whether internal short circuit anomalies have occurred in battery cells online.

[0023] In some scenarios, the rate of change of the cell data can be used to identify whether the cell has an internal short circuit. For example, the magnitude of the rate of change of voltage or temperature can be used to identify whether the cell has an internal short circuit. However, this method is prone to misjudgment because it only looks at the derivative at a single point, and it is sensitive to operating condition disturbances and cannot identify complex internal short circuit anomalies.

[0024] In other application scenarios, outlier detection algorithms can also be used to identify short-circuit anomalies within battery cells. For example, by analyzing the statistical distribution characteristics of battery cell data, or calculating whether the Mahalanobis distance between the tested data and normal samples exceeds a preset threshold, it can be determined whether an anomaly has occurred in the battery cell. However, this method has certain limitations. It is highly dependent on data features, ignores the temporal order information of the data, and suffers from an unstable outlier threshold.

[0025] To address the shortcomings of existing methods for detecting internal short-circuit anomalies in battery cells, such as heavy reliance on large datasets, high model complexity, and insufficient ability to identify subtle, gradual faults, this application proposes a method for detecting internal short-circuit anomalies in battery cells. This method collects multi-source time-series data (voltage, temperature, current, etc.) from the battery cell at different operating stages and performs standardization processing to construct a multi-dimensional time-varying feature trajectory reflecting the dynamic behavior of the battery cell. Based on this, dimensionality reduction and trajectory similarity analysis methods are used to calculate the consistency of the feature trajectories of the cell to be identified and the sample cells in the time dimension, monitoring the evolution direction and morphological shift of the feature trajectories. This allows for accurate identification of subtle differences in the nascent stage of an internal short circuit within the battery cell, enabling early prediction of potential faults.

[0026] Figure 1This diagram illustrates a flowchart of a cell-level short-circuit anomaly detection method provided in an embodiment of this application. The method is executed by an electronic device to detect whether multiple cells in an energy storage system have internal short-circuit anomalies. The electronic device can be a terminal device including one or more processors, such as an energy storage system, a touchscreen phone, a smartphone, a tablet computer, or other electronic devices. The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application; no limitation is made here. The one or more processors included in the electronic device can be processors of the same type, such as one or more CPUs; or they can be processors of different types, such as one or more CPUs and one or more ASICs; no limitation is made here. Figure 1 As shown, the method includes the following steps 110 to 170.

[0027] Step 110: Collect the voltage, current and temperature of each cell at N acquisition times, and determine the first N×5 matrix corresponding to each cell based on the collected voltage, current and temperature.

[0028] For ease of explanation, this application uses an energy storage system comprising 200 battery cells, with N being 100, as an example. At each of the 100 data acquisition moments, the voltage, current, and temperature of each of the 200 cells are collected, resulting in the voltage, current, and temperature data for each of the 200 cells. Specifically, the voltage, current, and temperature of each cell can be collected using voltage sensors, current sensors, and temperature sensors.

[0029] It is understandable that if 200 cells are connected in series in an energy storage system, the current of all 200 cells will be the same at the same sampling time. Therefore, in order to improve data acquisition efficiency, at each sampling time, only the current of any one cell can be collected as the current of all cells at that sampling time, thereby reducing the number of times the current sensor collects the current and improving data acquisition efficiency.

[0030] In this step, the first matrix corresponding to each battery cell includes N rows of data. These N rows are determined based on the voltage, current, and temperature of the battery cell collected at N time points. Specifically, if the collection time t is the r-th collection time among the N collection times, and the voltage, current, and temperature of the i-th battery cell collected at collection time t are respectively... , and In this step, The data in the r-th row of the first matrix corresponding to the i-th cell is determined. It is worth noting that the above explanation only uses the determination of the r-th row of the first matrix corresponding to the i-th cell as an example. The determination of the data in each row of the first matrix corresponding to other cells is the same as that of the i-th cell, and will not be repeated here.

[0031] Step 120: Normalize the first matrix corresponding to each cell to obtain the second matrix corresponding to each cell.

[0032] In this step, the data in each row of the first matrix is ​​normalized to obtain the second matrix. Taking the i-th cell as an example, the average value of each column of data in the first matrix corresponding to the i-th cell is first determined to obtain the average value matrix corresponding to the i-th cell. Then, the variance of each column of data is determined to obtain the variance matrix corresponding to the i-th cell. Next, the difference between the first matrix and the average value matrix of the i-th cell is determined to obtain the difference matrix. Finally, the ratio of the difference matrix to the variance matrix is ​​determined to obtain the ratio matrix, which is the second matrix corresponding to the i-th cell.

[0033] Step 130: Determine the N×5 third matrix based on the first matrix corresponding to all battery cells. .

[0034] In this step, the third matrix is ​​determined. A reference used to characterize a normal battery cell, so that subsequent steps can be based on the third matrix. Determine if any internal short circuit has occurred in each cell. This involves determining the third matrix. Taking the j-th row vector as an example, in this step, preferably, the third matrix is ​​obtained by taking the median of the elements at corresponding positions in the j-th row vector of the first matrix of all battery cells. The j-th row Vector. Specifically, determine the first median of all data in the j-th row and 1-th column of the first matrix of all cells. Determine the second median of all data in the j-th row and 2nd column of the first matrix for all battery cells. Determine the third median of all data in the j-th row and 3rd column of the first matrix for all battery cells. Determine the fourth median of all data in the j-th row and 4-th column of the first matrix for all battery cells. Determine the fifth median of all data in the j-th row and 5th column of the first matrix for all battery cells. Thus, the third matrix is ​​obtained. The j-th row vector , where the third matrix , j is a positive integer, and j≤N.

[0035] To determine the median of a set of data, arrange the data in ascending (or descending) order, and the value at the midpoint of the sequence is the median. If the sequence is odd, the median is equal to the nth position. The number of elements; if the sequence is even, the median is equal to the first element. and The average of the number.

[0036] For multiple cells in an energy storage system, the number of normal cells is usually greater than the number of abnormal cells. Therefore, in the first matrix corresponding to all cells determined in step 110, the same data point of different cells at the same time shows a clustered distribution characteristic centered on the data of normal cells. Therefore, in this step, determining the third matrix through median calculation can effectively eliminate outlier data caused by individual cell anomalies, avoid interference from extreme outliers on the benchmark data, and ensure that the third matrix can accurately reflect the real operating status and common characteristics of the normal cell group in the energy storage system. This improves the robustness and reliability of the benchmark reference vector, laying a stable and reliable data foundation for subsequent judgment on whether a cell has experienced an internal short circuit anomaly.

[0037] To further improve the accuracy of the third matrix, in some other embodiments, the third matrix... The j-th row The following method is used to determine the j-th row: For the elements at the corresponding positions in the first row vector of all battery cells, select the median and the values ​​adjacent to the median, and calculate the average of the selected data to obtain the j-th row. vector.

[0038] Step 140: For the third matrix Principal component analysis was performed to determine a 5×2 characteristic matrix used to characterize the cell state evolution. .

[0039] Specifically, the feature matrix can be determined through the following steps a1 to a5. .

[0040] Step a1: For the third matrix Perform mean-removal processing to obtain the mean-removed matrix. .

[0041] Specifically, the third matrix can be obtained by the following formulas (1) and (2). Perform mean removal processing.

[0042] (1)

[0043] (2)

[0044] Step a2: Determine the covariance matrix using the following formula (3). .

[0045] (3)

[0046] Step a3: Determine the covariance matrix First eigenvalue Second eigenvalue .

[0047] Among them, determining the covariance matrix eigenvalues The first eigenvalue is obtained in descending order. Second eigenvalue In this way, the first two principal components are retained to cover most of the variance, achieving dimensionality reduction to two-dimensional visualization.

[0048] Step a4: Determine the first eigenvalue The corresponding first characteristic matrix and the second eigenvalue The corresponding second characteristic matrix .

[0049] Step a5: [The first feature matrix is ​​then...] Second characteristic matrix The resulting matrix is ​​determined to be the characteristic matrix. .

[0050] in, and .

[0051] Step 150: For the i-th cell, use the following formula (4) to convert the i-th cell into its corresponding... The second matrix is ​​mapped to the characteristic matrix. The characteristic trajectory of the i-th cell is obtained. .

[0052] (4)

[0053] In step 150, only the i-th cell is used as an example. If the energy storage system includes m cells, then when i is 1, 2, ..., m, the characteristic trajectory corresponding to each cell is determined. Furthermore, this explanation also applies when the i-th cell is used as an example again in the following text, and will not be repeated hereafter.

[0054] In this application, the cell feature trajectory is determined by the above formula (4) to realize the reduction of multidimensional time-series features to two-dimensional trajectory projection through PCA. Its essence is to transform the "high-dimensional state evolution" of the cell during operation into a "two-dimensional geometric curve", thereby preserving the core change mode of the cell and simplifying the analysis process. When the cell is operating normally, its corresponding cell feature trajectory presents a stable form; when the cell experiences an internal short circuit anomaly, its features will shift, and its corresponding cell feature trajectory will be deformed due to the feature shift. Therefore, by monitoring the similarity between the cell feature trajectory and the cell feature trajectory corresponding to a normal cell, it is possible to detect whether the cell has experienced an internal short circuit anomaly, and it can be identified in the early stage of the cell experiencing an internal short circuit.

[0055] Step 160: Determine the trajectory evolution speed and trajectory evolution direction of the characteristic trajectory of each cell.

[0056] Specifically, taking the i-th cell as an example, the cell characteristic trajectory corresponding to the i-th cell can be determined by the following formula (5). trajectory evolution speed .

[0057] (5)

[0058] in, It is the difference vector formed by two adjacent points in the characteristic trajectory of the i-th cell. Let be the magnitude of the difference vector.

[0059] Determining the trajectory evolution rate is primarily aimed at capturing the degree of disruption to the "stable evolution law" of the cell's characteristic trajectory and using it as an early warning signal for internal short-circuit anomalies in the cell. Under normal operating conditions, the rate of co-evolution of the cell's voltage, temperature, current, and other characteristics is stable (e.g., the rate of slow temperature rise and the frequency of periodic voltage fluctuations remain consistent). Therefore, the trajectory evolution rate of the corresponding cell characteristic trajectory also remains constant or predictable (e.g., represented as an ellipse rotating with a fixed period in a two-dimensional projection). Thus, by calculating the trajectory evolution rate of the cell's corresponding characteristic trajectory, it is possible to subsequently determine whether an internal short-circuit anomaly has occurred in the cell.

[0060] The characteristic trajectory of the i-th cell can be determined by the following formula (6). The direction of trajectory evolution .

[0061] (6)

[0062] in, and These are the difference vectors formed by two adjacent points in the characteristic trajectory of the i-th cell.

[0063] The "directivity" of the co-evolution of features is determined by the internal laws of the battery cell. During normal operation, the dynamic changes of features such as voltage, temperature, and current in the battery cell follow the inherent electrochemical coupling law. However, when an internal short circuit occurs in the battery cell, it leads to a local micro-cell effect (direct contact between the positive and negative electrodes), thereby generating additional short-circuit current Isc and heat Qsc, which will disrupt the original feature coupling relationship.

[0064] The evolution direction of the cell's characteristic trajectory can be quantified by calculating the difference vector of the trajectory points or the angle of the principal component projection. During normal cell operation, the evolution direction of the cell's characteristic trajectory aligns with the physical meaning represented by the principal component. However, when an internal short circuit occurs in the cell, the mean value of its direction angle undergoes a sudden change (or even a reversal), indicating that the characteristic coupling relationship represented by the principal component has been disrupted. Therefore, in this step, the trajectory evolution direction of the cell's characteristic trajectory is calculated so that subsequent judgments on whether the cell has experienced an anomaly can be made based on the trajectory evolution direction.

[0065] Step 170: Determine whether an internal short circuit anomaly has occurred in each cell based on the trajectory evolution speed and trajectory evolution direction of the cell characteristic trajectory corresponding to each cell.

[0066] In this step, to improve the efficiency of detecting whether an internal short circuit anomaly has occurred in a battery cell, a trajectory evolution speed threshold and a trajectory evolution direction threshold can be set. The trajectory evolution speed and trajectory evolution direction of the characteristic trajectory corresponding to each battery cell are compared with the corresponding thresholds, and then the internal short circuit anomaly of each battery cell is determined based on the comparison results. For example, if the trajectory evolution speed of the characteristic trajectory corresponding to a certain battery cell is greater than the trajectory evolution speed threshold and the trajectory evolution direction is greater than the trajectory evolution direction threshold, or if the trajectory evolution speed of the characteristic trajectory corresponding to a certain battery cell is less than or equal to the trajectory evolution speed threshold and the trajectory evolution direction is greater than the trajectory evolution direction threshold, then it is determined that the battery cell has an internal short circuit anomaly.

[0067] It is worth noting that in order to identify whether a cell has experienced a minor internal short circuit, that is, to determine the level of the internal short circuit, different levels of trajectory evolution speed thresholds and trajectory evolution direction thresholds can be set. When it is determined which level of threshold the trajectory evolution speed and trajectory evolution direction of the cell's characteristic trajectory belong to, the abnormality level of the cell can be determined.

[0068] In this embodiment, a first matrix is ​​determined based on the voltage, temperature, and current of the battery cells collected at multiple times to extract multi-dimensional features that change over time. Each first matrix is ​​then normalized and time-aligned to form a time-varying feature vector sequence, ensuring comparability between different battery cells in the time domain and providing a unified scale for determining whether a battery cell has experienced an internal short-circuit anomaly based on its corresponding feature trajectory. Next, a third matrix is ​​determined based on the first matrices of all battery cells, serving as the sample matrix for normal battery cells. To determine the difference between the second matrix corresponding to each battery cell and this sample matrix, and thus whether an internal short-circuit anomaly has occurred in each battery cell, this application performs principal component analysis on the sample matrix corresponding to normal battery cells to obtain the principal component basis matrix (i.e., feature matrix P). The high-dimensional time-varying features (i.e., the second matrix) corresponding to each battery cell are projected into a low-dimensional space to form a two-dimensional trajectory (i.e., the battery cell feature trajectory), thereby achieving information compression and noise suppression, making the changes in the battery cell feature trajectory better reflect the potential abnormal trends of the battery cell. Then, based on the motion characteristics of the cell feature trajectory in low-dimensional space, the local trajectory evolution rate of the cell feature trajectory is calculated. With the direction of trajectory evolution Thus, the trajectory evolution speed of the cell's characteristic trajectory can be observed. With the direction of trajectory evolution This study reveals the dynamic changes in the characteristic trajectory of battery cells, such as deflection and fluctuation, and further analyzes the trajectory evolution rate based on the characteristic trajectory of battery cells. With the direction of trajectory evolution When determining whether a battery cell has an internal short circuit, this method can improve the accuracy of the judgment results compared to methods that rely on a single parameter (such as the cell's voltage, current, or temperature).

[0069] In some embodiments, in order to improve the accuracy of determining whether an internal short circuit has occurred in a battery cell, the following steps b1 to b4 are used to determine whether an internal short circuit has occurred in a battery cell.

[0070] Step b1: Determine the characteristic trajectory of the reference cell using the following formula (7). .

[0071] (7)

[0072] After determining the cell feature trajectory corresponding to each cell through the above formula (4), the reference cell feature trajectory is determined through the above formula (7) in this application. Subsequently, the reference cell feature trajectory can be used as the cell feature trajectory corresponding to the normal cell. Then, based on the cell feature trajectory corresponding to each cell and the reference cell feature trajectory, it can be determined whether each cell has an internal short circuit abnormality.

[0073] Step b2: For the i-th cell, determine the cell characteristic trajectory corresponding to the i-th cell. Compared with the characteristic trajectory of the reference cell DTW distance between ,in, .

[0074] In this step, the DTW distance between the characteristic trajectory of each cell and the characteristic trajectory of the reference cell is calculated in order to quantify the similarity between the characteristic trajectory of each cell and the characteristic trajectory of the reference cell, so as to determine whether the characteristic trajectory of each cell has been significantly deviated due to internal short circuit, and thus determine whether the cell has experienced internal short circuit anomaly.

[0075] To simplify the explanation of the DTW distance calculation steps, the following illustration uses a one-dimensional point sequence. In practical applications, the cell feature trajectory is a two-dimensional point sequence, but the calculation steps for the DTW distance are similar. Specifically, to explain how to determine the DTW distance between two trajectories, we will use a one-dimensional point sequence as an example. , Let's take an example. Matrix D is determined using the following formula (8).

[0076] (8)

[0077] Where p and q are both natural numbers. Using the above formula, the matrix can be obtained. .

[0078] Next, initialize the cumulative cost matrix C and fill in the table according to dynamic programming. Let C and D have the same dimension, and determine the matrix C according to the recursive relationship of the following formulas (9) to (11).

[0079] (9)

[0080] (10)

[0081] (11)

[0082] Based on the above recursive relationship, the specific data in the first row of matrix C can be determined as follows: ; ; ; ; =20. And the data in the first column of matrix C is specifically as follows: ; ; ; .

[0083] The data in matrix C, excluding the first row and first column, are determined specifically by the formula (11) above. ; ; ; .

[0084] Therefore, matrix The final cumulative minimum cost is C(5,5)=8.

[0085] Next, the optimal alignment path is obtained by backtracking from the endpoint C(5,5) back to the starting point C(1,1). At each step, the smallest adjacent point is selected. If the values ​​are the same, the diagonal point is selected first to obtain the backtracking path. For the path C(5,5)→C(5,4)→C(4,3)→C(3,2)→C(2,1)→C(1,1), the length of the path is calculated to be L=6. The normalized DTW distance can then be determined by the following formula (12).

[0086] (12)

[0087] The above is an example of DTW calculation for a one-dimensional sequence. For the two-dimensional cell feature trajectory in the embodiments of this application, the Euclidean distance calculation formula should be used in formula (8), and the remaining steps are similar.

[0088] Step b3: Based on the DTW distance corresponding to the i-th cell Determine the similarity between the cell feature trajectory corresponding to the i-th cell and the reference cell feature trajectory. .

[0089] Specifically, the similarity of the i-th cell can be determined by the following formula (13). .

[0090] (13)

[0091] in, These are preset parameters, for example, they can be 0.4 or 0.5.

[0092] Step b4: Based on the similarity of the i-th cell The trajectory evolution speed and direction of the characteristic trajectory of the battery cell are used to determine whether the i-th battery cell has an internal short circuit anomaly.

[0093] Specifically, this step can be achieved through the following steps b41 to b43.

[0094] Step b41: For the i-th cell, determine the comprehensive characteristic change rate corresponding to the i-th cell using the following formula (14). .

[0095] (14)

[0096] in, This is a preset parameter, for example, 0.4 or 0.5.

[0097] Step b42: Determine the overall anomaly degree corresponding to the i-th cell using the following formula (15). .

[0098] (15)

[0099] in, and For example, to preset weights, It is 0.7. It is 0.3 or, It is 0.6. It is 0.4.

[0100] Step b43: If the overall anomaly degree corresponding to the i-th cell is If the value exceeds a preset threshold, it is determined that the i-th cell has experienced an internal short circuit anomaly.

[0101] The preset threshold can be set as needed, such as 0.5 or 0.6.

[0102] To identify whether a micro-internal short circuit anomaly has occurred in a battery cell, in this embodiment of the application, preferably, two preset thresholds are pre-set, namely a first preset threshold and a second preset threshold, wherein the second preset threshold is greater than the first preset threshold. If the comprehensive anomaly degree corresponding to the i-th battery cell is... If the overall anomaly level of the i-th cell exceeds the first preset threshold, the level of the internal short circuit anomaly is determined to be Level 1; if the overall anomaly level of the i-th cell is... If the value is greater than the second preset threshold, the level of the internal short circuit anomaly in the i-th cell is determined to be the second level, where the second level is higher than the first level.

[0103] Figure 2 This illustration shows schematic diagrams of the characteristic trajectories of a reference cell, a normal cell, a cell with a minor internal short circuit, and a cell with a severe internal short circuit, respectively, provided in embodiments of this application. Figure 2 As shown, Figure 2 (a) includes the normal trajectory (i.e., the characteristic trajectory of the cell corresponding to the normal cell) and the normal reference curve (i.e., the characteristic trajectory of the reference cell). As can be seen from (a), the normal trajectory (green solid line) closely surrounds and fits the normal reference curve (blue dashed line). The two are highly consistent in shape, indicating that the operating state of the cell corresponding to the green solid line is stable and conforms to the reference mode. Therefore, it can be determined that the cell characteristic trajectory is the cell corresponding to the green solid line, which is the normal cell.

[0104] Figure 2(b) includes a slight internal short circuit trajectory (i.e., the characteristic trajectory of the cell corresponding to the cell with a slight internal short circuit) and a normal reference curve. As can be seen from (b), although the shape of the slight internal short circuit trajectory (orange solid line) still roughly maintains an elliptical outline, it has shown obvious deviation and fluctuation. It deviates from the normal reference curve (blue dashed line) to a certain extent, and this deviation is greater than the deviation between the normal trajectory and the normal reference curve in (a). This reflects that the cell corresponding to the orange solid line has a slight internal short circuit abnormality, but has not yet reached the level of a serious internal short circuit abnormality. Therefore, it can be determined that the cell characteristic trajectory is the cell corresponding to the orange solid line, which is the cell with a slight internal short circuit. Figure 2 (c) includes the severe internal short circuit trajectory (i.e., the characteristic trajectory of the cell corresponding to the cell that has experienced a severe internal short circuit) and the normal reference curve. As can be seen from (c), the shape of the severe internal short circuit trajectory (red solid line) is completely disordered, showing violent and irregular fluctuations, and the deviation from the normal reference curve is extremely large. This intuitively reflects the violent state change of the cell under severe fault. Therefore, it can be determined that the cell corresponding to the red solid line is the cell that has experienced a severe internal short circuit.

[0105] Depend on Figure 2 It can be seen that by comparing the cell characteristic trajectories corresponding to the cells under these three different states, it is clear that as the degree of internal short circuit fault intensifies, the deviation of the cell characteristic trajectory from the normal reference curve becomes more and more significant. This provides an intuitive and effective basis for internal short circuit anomaly detection and fault degree assessment based on trajectory morphology.

[0106] In other words, in this embodiment, the method of determining whether an internal short circuit anomaly has occurred in each battery cell is achieved by comparing the characteristic trajectory of each battery cell with that of a reference battery cell. This method has the advantages of clear physical meaning (i.e., if the trajectory deviation is large, the corresponding battery cell has an internal short circuit anomaly), high computational efficiency (i.e., fast matching speed of two-dimensional trajectories), and strong interpretability (i.e., graphical results are easy to verify). Furthermore, dynamic time warping (DTW) is performed on the real-time sampled characteristic trajectory of the battery cell and the reference battery cell characteristic trajectory. By recursively deriving the minimum matching cost matrix, the minimum matching distance between the characteristic trajectories of the battery cells is obtained, and this minimum matching distance is used to measure the degree of deviation of the battery cell from the normal mode, thereby effectively adapting to situations with different timing lengths and operating rates.

[0107] Furthermore, to enhance the robustness of calculating the similarity between the characteristic trajectory of each battery cell and the characteristic trajectory of a reference battery cell, this application uses the trajectory evolution speed as a weighting factor for time alignment when employing algorithms such as Dynamic Time Warping (DTW) for similarity prediction. Specifically, the evolution speed of normal trajectories is relatively stable, so the influence of time scaling can be ignored when calculating similarity; while for trajectories with abnormal evolution speeds, by adjusting the corresponding speed weights, it is possible to more accurately identify those abnormal trajectories that are "similar in shape but different in speed" (such as the characteristic trajectory of a battery cell corresponding to a "slow divergence" cell in the early stage of an internal short circuit). Furthermore, by calculating the DTW distance and using it to determine whether an internal short circuit anomaly has occurred in the cell, variations in evolution speed can be tolerated. That is, an internal short circuit in the cell may cause local acceleration (e.g., high speed in the divergence segment) or deceleration (e.g., slow initial offset) in the cell's characteristic trajectory. The DTW algorithm allows for "stretching" (setting more alignment points in the slow segment) or "compression" (setting fewer alignment points in the fast segment) on the time axis, thus focusing on the shape of the trajectory rather than its evolution speed. Moreover, the DTW distance accurately captures the offset of the cell's characteristic trajectory shape. By finding the optimal path, DTW aligns the first half of the normal elliptical trajectory with the overlapping portion of the current trajectory, and the second half with the divergence segment. This significantly increases the accumulated distance, thus correctly reflecting the abnormal state of the cell.

[0108] In summary, compared with existing cell fault diagnosis methods based on voltage change rate, outlier detection, or neural network models, the cell short-circuit anomaly detection method provided in this application achieves highly sensitive detection of the early evolution process of cell short circuits by constructing multidimensional time-varying cell feature trajectories and introducing trajectory similarity analysis. Furthermore, this application extracts the trajectory evolution features of signals over time based on the time-series changes of multi-source signals such as voltage, temperature, and current of the cell. This reflects the dynamic behavior of the cell during operation, rather than relying solely on static instantaneous values, thus significantly improving the accuracy of anomaly identification and enabling early detection of cell anomalies. Moreover, this application achieves feature dimensionality reduction and trajectory projection through principal component analysis, reducing computational load and noise interference. The extracted low-dimensional trajectories have clear physical meanings, facilitating embedding and maintenance in a Battery Management System (BMS). Simultaneously, by using trajectory similarity (such as DTW distance) to measure differences between cells, individuals deviating from the normal evolutionary feature trajectories within a group can be identified, avoiding misjudgment of individual cell anomalies and achieving more stable group consistency analysis. Furthermore, by comprehensively monitoring trajectory similarity and evolution rate changes, this application can output early warning signals at the early stages of micro-internal short circuits or abnormal electrode contact in the battery cell, possessing online detection and predictive maintenance capabilities. In summary, this application has advantages such as high sensitivity, high reliability, low computational complexity, and strong interpretability, making it suitable for BMS online diagnostics and health management scenarios in energy storage and power battery systems, effectively improving system safety and lifespan.

[0109] Figure 3 The diagram shows a structural schematic of the cell short-circuit anomaly detection device provided in an embodiment of this application. The specific embodiments of this application do not limit the specific implementation of the cell short-circuit anomaly detection device.

[0110] like Figure 3 As shown, the short circuit fault detection device 200 in the battery cell may include a processor 202 and a memory 204.

[0111] The memory 204 is used to store the computer program 206. The memory 204 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive. The computer program 206 may include computer-executable instructions.

[0112] The processor 202 is used to execute the computer program 206 to implement the above-described embodiment of the method for detecting short circuit anomalies within a battery cell.

[0113] The processor 202 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The one or more processors included in the cell short-circuit fault detection device 200 may be processors of the same type, such as one or more CPUs; or they may be processors of different types, such as one or more CPUs and one or more ASICs.

[0114] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for detecting short circuits within a battery cell.

[0115] This application provides a computer program that can be executed by a processor to implement the above-described method for detecting short circuit anomalies within a battery cell.

[0116] This application provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described embodiment of the cell short-circuit anomaly detection method.

[0117] In the several embodiments provided in this application, any function, if implemented as a software functional module / unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the technical solution of this application can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or other electronic device) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing computer program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0118] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of this application are not directed to any particular programming language. It should be understood that the content of this application described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of this application.

[0119] It should be noted that the above embodiments are illustrative of this application and not restrictive, and those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In claims enumerating several means, several units or modules of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

[0120] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting short-circuit anomalies within a battery cell, used to detect multiple battery cells in an energy storage system, characterized in that, The method includes: The voltage, current, and temperature of each battery cell are collected at N acquisition times. Based on the collected voltage, current, and temperature, an N×5 first matrix is ​​determined for each battery cell, where each row of the first matrix corresponding to the i-th battery cell is... , , and These are the voltage, current, and temperature of the i-th cell collected at time t, respectively. Different rows of data correspond to different collection times t. N and i are both positive integers, and N >

2. Normalize the first matrix corresponding to each battery cell to obtain the second matrix corresponding to each battery cell; The third matrix (N×5) is determined based on the first matrix corresponding to all battery cells. ,in, The j-th row of the third matrix , Each data point is determined based on the j-th row of all data in the first matrix, where j is a positive integer and j≤N; For the third matrix Principal component analysis was performed to determine a 5×2 characteristic matrix used to characterize the cell state evolution. ; Through formula Map the second matrix of each battery cell to the feature matrix. This yields the characteristic trajectory of each battery cell, where... The cell feature trajectory corresponding to the i-th cell This is the second matrix corresponding to the i-th battery cell; The trajectory evolution speed and trajectory evolution direction of the characteristic trajectory of each battery cell are determined respectively; Based on the trajectory evolution speed and direction of the characteristic trajectory corresponding to each cell, it is determined whether an internal short circuit anomaly has occurred in each cell; wherein, The third matrix Principal component analysis was performed to determine a 5×2 characteristic matrix used to characterize the cell state evolution. ,include: For the third matrix Perform mean-removal processing to obtain the mean-removed matrix. ; Through formula Determine the covariance matrix ; Determine the covariance matrix First eigenvalue Second eigenvalue ; Determine the first feature value The corresponding first characteristic matrix and the second eigenvalue The corresponding second characteristic matrix ; The first feature matrix and the second feature matrix The constructed matrix is ​​determined as the characteristic matrix. ,in, .

2. The method according to claim 1, characterized in that, The third matrix The j-th row It is obtained by taking the median of the elements at the corresponding positions in the j-th row vector of the first matrix for all battery cells.

3. The method according to claim 1, characterized in that, Determine the trajectory evolution rate of the characteristic trajectory of each battery cell, including: For the i-th cell, the formula is used. Determine the trajectory evolution rate of the characteristic trajectory of the i-th cell. ,in, It is the difference vector formed by two adjacent points in the characteristic trajectory of the i-th cell.

4. The method according to claim 1, characterized in that, Determine the trajectory evolution direction of the characteristic trajectory of each battery cell, including: Through formula Determine the trajectory evolution direction of the characteristic trajectory of the i-th cell. ,in, and These are the difference vectors formed by two adjacent points in the characteristic trajectory of the i-th cell.

5. The method according to claim 1, characterized in that, The step of determining whether each battery cell has experienced an internal short circuit anomaly based on the trajectory evolution speed and trajectory evolution direction of the characteristic trajectory corresponding to each battery cell includes: Through formula Determine the characteristic trajectory of the reference cell ; For the i-th cell, determine the cell feature trajectory corresponding to the i-th cell. With the reference cell characteristic trajectory DTW distance between ; According to the DTW distance corresponding to the i-th cell Determine the similarity between the feature trajectory of the i-th cell and the feature trajectory of the reference cell. ; Based on the similarity corresponding to the i-th cell The trajectory evolution speed and direction of the characteristic trajectory of the battery cell are used to determine whether the i-th battery cell has an internal short circuit anomaly.

6. The method according to claim 5, characterized in that, The DTW distance corresponding to the i-th cell Determine the similarity between the feature trajectory of the i-th cell and the feature trajectory of the reference cell. ,include: Through formula Determine the similarity corresponding to the i-th cell. ,in, These are preset parameters; The similarity based on the i-th cell The trajectory evolution speed and direction of the cell characteristic trajectory are used to determine whether the i-th cell has an internal short circuit anomaly, including: For the i-th cell, the formula is used. Determine the rate of change of comprehensive characteristics corresponding to the i-th cell. ,in, These are preset parameters. Let be the trajectory evolution speed of the characteristic trajectory of the i-th cell. The trajectory evolution direction of the characteristic trajectory of the i-th cell. ; Through formula Determine the overall anomaly degree corresponding to the i-th cell. ,in, and Preset weights; If the overall anomaly degree corresponding to the i-th cell If the value exceeds a preset threshold, it is determined that the i-th cell has experienced an internal short circuit anomaly.

7. The method according to claim 6, characterized in that, If the comprehensive anomaly degree corresponding to the i-th cell If the value exceeds a preset threshold, an internal short circuit anomaly is determined in the i-th cell, including: If the overall anomaly degree corresponding to the i-th cell If the value is greater than the first preset threshold, the level of the internal short circuit anomaly in the i-th cell is determined to be the first level; If the overall anomaly degree corresponding to the i-th cell If the value is greater than the second preset threshold, the level of the internal short circuit anomaly in the i-th cell is determined to be the second level, wherein the second preset threshold is greater than the first preset threshold, and the second level is higher than the first level.

8. A short-circuit fault detection device for a battery cell, comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the cell short-circuit anomaly detection method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the cell short-circuit anomaly detection method according to any one of claims 1 to 7.

Images