Battery fault early warning method based on improved clustering analysis

Abstract

The invention discloses a battery fault early warning method based on improved clustering analysis, which finds a balance between the accuracy and the efficiency of battery fault early warning by constraining the setting condition of a scanning radius and dynamically selecting a reference cluster, realizes accurate detection and timely early warning of a battery fault, and improves the early warning efficiency of the battery fault. And stable operation and use safety of the battery system are effectively ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of battery fault detection, and in particular to a battery fault early warning method based on improved cluster analysis. Background Art

[0002] Under extreme operating conditions, lithium-ion batteries may experience thermal runaway reactions due to faults such as overcharging and short circuits, posing serious safety risks. Due to the complex chemical reactions within the battery and the uncertainty of external environmental conditions, it is difficult to accurately diagnose abnormal battery conditions and potential faults. Existing methods typically attempt to promptly detect and prevent potentially dangerous conditions by real-time monitoring of multi-dimensional sensing parameters such as battery voltage and temperature, and analyzing the collected data. However, the accuracy of fault detection is less than ideal, and it is difficult to ensure the timeliness of fault detection. Therefore, the field is looking forward to a method that can accurately detect battery faults and provide timely warnings to ensure stable operation and safe use of the battery system. Summary of the Invention

[0003] The present invention aims to accurately detect battery failures and issue early warnings in a timely manner to ensure stable operation and safe use of the battery system, and provides a battery failure early warning method based on improved cluster analysis.

[0004] To achieve this object, the present invention adopts the following technical solutions: A battery failure early warning method based on improved cluster analysis is provided, comprising the steps of: S1, real-time collection of status data of each single cell in the battery pack, including single cell Battery voltage ,temperature , deformation pressure , gas concentration One or more of; Indicates the current time; S2, clustering the single cells based on the status data, and then dynamically selecting a reference cluster; S3, based on the benchmark clustering, calculating the maximum similarity distance between each cluster and performing threshold comparison, and then outputting the fault warning classification result for each cluster.

[0005] Preferably, in step S2, after normalizing the state data associated with each single cell, multi-dimensional characteristic time series data associated with each single cell is obtained. The normalization method is expressed by the following formula (1): (1) In formula (1), Represents the normalized value; Indicates one of the state data; Respectively are the minimum and maximum values ​​of the same state data within the acquisition time window.

[0006] Preferably, in step S2, the method for clustering the single cells comprises the steps of: A1, calculate the weighted similarity distance between single cells , Indicates single battery and single battery The weighted similarity distance between them; A2, set the scanning radius , Single battery and single battery Add to the matching set; A3, calculate single battery The weighted similarity distance of each single battery in the matching set is determined to be less than or equal to the scanning radius. , If so, the single cell Add to the matching set; A4. After all the single cells have been judged as above, the remaining single cells are added to the non-compliant set.

[0007] Preferably, For single battery and single battery Voltage similarity distance between , temperature similarity distance , deformation pressure similarity distance , gas concentration similarity distance The weighted average of one or more of .

[0008] Preferably, the single battery and single battery The similarity distance of the same state data The calculation method is expressed by the following formula (2): (2) Indicates the minimum path The value of the node; represents the total number of nodes in the minimum path; The minimum path is found by the following method steps: C1, for calculation Construct the corresponding matrix; C2, finding a path from the element in the top right corner of the matrix to the element in the bottom left corner of the matrix with the smallest sum of element values as the minimum path found.

[0009] Preferably, in step C1, the calculation is The method for constructing the corresponding matrix is expressed by formula (3) as follows: (3) denotes the value of the element in the i-th column and the j-th row of the matrix; denotes the value of the element in the i-th column and the j-th row of the matrix; denotes the value of the element in the i-th column and the j-th row of the matrix. denotes the i-th state data value in the first state data time series collected in the same type of state data of the single battery at each time in the collection time window, and the elements in the first state data time series are arranged in chronological order; denotes the i-th state data value in the second state data time series collected in the same type of state data of the single battery at each time in the same collection time window of the state data of the single battery, and the elements in the second state data time series are arranged in chronological order; denotes the i-th state data value in the second state data time series collected in the same type of state data of the single battery at each time in the same collection time window of the state data of the single battery, and the elements in the second state data time series are arranged in chronological order; denotes the i-th state data value in the second state data time series collected in the same type of state data of the single battery at each time in the same collection time window of the state data of the single battery, and the elements in the second state data time series are arranged in chronological order; denotes the value of the element in the i-th column and the j-th row of the matrix; denotes the value of the element in the i-th column and the j-th row of the matrix; denotes the value of the element in the i-th column and the j-th row of the matrix; denotes the value of the element in the i-th column and the j-th row of the matrix; denotes the value of the element in the i-th column and the j-th row of the matrix. denotes the value of the element in the i-th column and the j-th row of the matrix. Preferably, in step C2, the constraint condition for finding the minimum path is:

[0010] with the element in the top right corner of the matrix as the starting point of the path and the element in the bottom left corner of the matrix as the ending point of the path, and with the minimum path formed finally.

[0011] Preferably, in step S2, the method for dynamically selecting the reference cluster includes the following steps: S21, judging whether the clustering result of each single battery is that the number of the non-conforming set is at least 1 and the number of the conforming set is 0,​​​​​​​​ If yes, an abnormal alarm is issued for the battery pack; If not, go to step S22; S22, determining whether the number of the matching set is 1 and the number of the non-matching set is 0, If yes, the battery pack is determined to be normal and the fault warning process is terminated; If not, the only matching set or the matching set with the largest number of single cells is selected as the reference cluster, and the remaining matching sets or non-matching sets are each a remaining cluster.

[0012] Preferably, step S3 specifically includes the steps of: S31, calculating the inter-cluster similarity distance between each remaining cluster except the benchmark cluster and the benchmark cluster; S32, setting alarm threshold intervals corresponding to different fault warning levels; S33, judging whether the inter-class similarity distance falls within the corresponding alarm threshold range, If yes, classify the remaining clusters into the fault warning level corresponding to the alarm threshold interval they fall into and issue an alarm; If not, it is determined that each battery cell in the remaining cluster operates normally.

[0013] Preferably, in step S31, the method for calculating the inter-class similarity distance is: The similarity distance between each single battery in the same remaining cluster and each single battery in the reference cluster is calculated, and the maximum similarity distance calculated is used as the inter-cluster similarity distance between the remaining cluster and the reference cluster.

[0014] This application constrains the scanning radius By dynamically selecting benchmark clusters and ensuring the balance between the accuracy and efficiency of battery fault warning, the system can accurately detect battery faults and provide timely warnings, effectively ensuring the stable operation and safe use of the battery system. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0016] Figure 1 This is a diagram of the steps for implementing the battery failure early warning method based on improved cluster analysis provided in an embodiment of the present application; Figure 2It is a single battery and single battery An example diagram of a matrix constructed from time series collected for the same state data at different times within the same acquisition time window; Figure 3 is an example graph of the minimum path found in the matrix; DETAILED DESCRIPTION

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings and through specific implementation methods.

[0018] Among them, the drawings are only used for illustrative purposes and represent only schematic diagrams rather than actual pictures, and should not be understood as limiting this patent; in order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0019] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "inside", "outside" and the like indicate an orientation or position relationship based on the orientation or position relationship shown in the drawings, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting this patent. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0020] In the description of the present invention, unless otherwise expressly specified or limited, when the term "connection" or the like appears to indicate a connection relationship between components, such term should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be internal communication between two components or an interaction between two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood in specific circumstances.

[0021] This embodiment of the application analyzes multidimensional time series data of sensor parameters during lithium-ion battery operation and combines it with threshold analysis of the maximum similarity distance between clusters to achieve multi-level fault warning, providing a basis for preventive maintenance. Furthermore, the battery fault warning method based on improved cluster analysis provided in this embodiment is integrated into the battery management system to enhance intelligent battery monitoring and ensure stable operation and safe use of the battery system.

[0022] The battery failure early warning method based on improved cluster analysis provided in this embodiment is as follows: Figure 1 As shown, the steps include: S1, real-time collection of status data of each single cell in the battery pack, including battery voltage ,temperature , deformation pressure , gas concentration , Indicates the current time, Indicates single battery , , Indicates the length of the sampling time window, such as , each moment corresponds to 1 minute, which means that sampling is done once every 1 minute, for a total of 5 times. , is the number of single cells in the battery pack; If a 1:1 ratio of sensors (such as temperature sensors, strain gauges, and gas concentration sensors) to battery cells is not achieved, the battery temperature data within the sensor's detection range will be the same. For example, if the ratio of temperature sensors to battery cells is 1:2, then the battery temperature data collected at the same time for two battery cells within the temperature sensor's detection range will be the same.

[0023] S2, clustering each single battery based on the status data of each single battery, and then dynamically selecting a benchmark cluster; In this embodiment, considering the inconsistent dimensions of different types of sensor data, the Min-Max method is used to normalize the state data to obtain normalized multi-dimensional feature time series data. The Min-Max method is expressed by the following formula (1): (1) In formula (1), Represents the normalized value; Indicates one type of state data; Respectively It is the minimum and maximum value of the same state data within the acquisition time window.

[0024] For example, for a single battery ,exist The single cell voltage collected within the sampling time window of length ,but , , then the expression for normalizing the single cell voltage of 3.0 by formula (1) is: .

[0025] Single battery after normalization of status data exist Multidimensional feature time series of moments , for 、 、 、 The corresponding normalized values.

[0026] In this embodiment, the method for clustering individual cells includes the following steps: A1, calculate the weighted similarity distance between single cells , Indicates single battery and single battery The weighted similarity distance between them; In this embodiment, For single battery and single battery Voltage similarity distance between , temperature similarity distance , deformation pressure similarity distance , gas concentration similarity distance The weighted average of one or more of 、 、 The weighted average of .

[0027] Preferably, , that is, the weights are equal to simplify the calculation difficulty.

[0028] Single battery and single battery The similarity distance between the same state data (like ) is calculated by the following formula (2): (2) Indicates the minimum path The value of the node; Indicates the total number of nodes in the minimum path; In this embodiment, It is a method to measure the similarity distance between two time series. For example, suppose that a single battery Battery voltage time series collected within the acquisition time window Expressed as: , Indicates the number of cells per moment within the acquisition time window. The voltage collected separately; single battery Battery voltage time series collected within the same acquisition time window Expressed as: , Indicates the number of cells per moment within the acquisition time window. The voltages collected respectively.

[0029] Assumptions , , then construct a time series and time series Composed of Figure 2 The matrix shown.

[0030] From the above explanation, we can see that the shortest path is found through the following method steps: C1, for calculation Construct the corresponding matrix. The construction method is expressed by the following formula (3): (3) Indicates the first Liedi The value of the row's elements; Indicates the number of cells per moment within the acquisition time window. The same type of state data is collected into the first state data time series state data values, and the elements in the first state data time series are arranged in the order of collection time; Indicates that the single battery In the same acquisition time window of the collected status data, for the same type of status data, the single battery The first time series of the second state data collected state data values, and the elements in the second state data time series are arranged in the order of collection time; Indicates the first Column, No. The element value of the row; Indicates the first Column, No. The element value of the row; Indicates the first Column, No. The element value of the row.

[0031] For example, Figure 2 In this embodiment, the next node of the current node is As a constraint condition, the element in the upper right corner of the matrix is taken as the starting point of the path, and the element in the lower left corner is taken as the ending point of the path, and finally the minimum path is formed.

[0032]

[0033] In this embodiment, the next node of the current node is As a constraint condition, the element in the upper right corner of the matrix is taken as the starting point of the path, and the element in the lower left corner is taken as the ending point of the path, and finally the minimum path is formed.

[0034] For example, Figure 3 In this embodiment, the next node of the current node is As a constraint condition, the element in the upper right corner of the matrix is taken as the starting point of the path, and the element in the lower left corner is taken as the ending point of the path, and finally the minimum path is formed.

[0035] After the calculation of the weighted similarity distance between the single batteries is completed, the method for clustering the single batteries in this embodiment proceeds to step: A2, the scanning radius is set, and the single battery and the single battery are added to the conforming set. A3, the weighted similarity distance between the single battery and each single battery in the conforming set is calculated, and it is determined whether they are all less than or equal to the scanning radius . If yes, the single battery is added to the conforming set. A4, after the above determination is made for all the single batteries, the remaining single batteries are added to the non-conforming set. Suppose that

[0036] , , , , , , , is preferred. Suppose the scanning radius , then the single battery and the single battery are added to the non-conforming set.

[0037] In step S2 of this embodiment, the method for dynamically selecting the reference clustering on which step S3 depends includes the following steps: S21, it is determined whether the clustering result of each single battery is that the number of the non-conforming set is at least 1 and the number of the conforming set is 0, If yes, an abnormal alarm is given to the battery pack. If not, go to step S22; S22, determine whether the number of matching sets is 1 and the number of non-matching sets is 0, If so, the battery pack is determined to be normal and the fault warning process is terminated; If not, the only matching set or the matching set with the largest number of single cells is selected as the benchmark cluster, and the remaining matching sets or non-matching sets are each a remaining set.

[0038] After dynamically selecting the benchmark cluster through step S2, Figure 1 As shown, the battery failure early warning method based on improved cluster analysis provided in this embodiment proceeds to the following steps: S3, based on the benchmark clustering, calculates the maximum similarity distance between each cluster and performs threshold comparison, and then outputs the fault warning classification result for each cluster.

[0039] Specifically, step S3 includes the steps of: S31, calculate the inter-class similarity distance between each remaining cluster except the benchmark cluster and the benchmark cluster, and the calculation method is: Calculate the similarity distance between each battery in the same remaining cluster and each battery in the benchmark cluster, and use the maximum similarity distance calculated as the inter-class similarity distance. The calculation method of the similarity distance between two batteries is the same as The calculation method of is not described in detail.

[0040] S32, setting alarm threshold intervals corresponding to different fault warning levels; For example, three alarm threshold intervals are set, namely [0, err1), [err1, err2), [err2, err3], and the corresponding fault warning levels are mild, moderate and severe respectively.

[0041] S33, determine whether the inter-class similarity distance falls within the corresponding alarm threshold range, If yes, the remaining clusters are classified into the fault warning level corresponding to the alarm threshold interval they fall into and an alarm is issued; If not, it is determined that each battery cell in the remaining cluster is operating normally.

[0042] For example, when the inter-cluster similarity distance falls within the [err1, err2) alarm threshold interval, it is determined that each single battery in the remaining cluster has a moderate fault.

[0043] In summary, this application constrains the scanning radius By dynamically selecting benchmark clusters and ensuring the balance between the accuracy and efficiency of battery fault warning, the system can accurately detect battery faults and provide timely warnings, effectively ensuring the stable operation and safe use of the battery system.

[0044] It should be noted that the above-mentioned detailed description is only the preferred embodiment of the present application and the applied technical principles. Those skilled in the art should understand that various modifications, equivalent replacements, changes and the like can be made to the present application. However, as long as these changes do not deviate from the spirit of the present application, they should be within the protection scope of the present application. In addition, some terms used in the present application specification and claims are not limited, but only for the convenience of description.

Claims

1. A battery failure early warning method based on improved cluster analysis, characterized in that: Including steps: S1, real-time collection of status data of each single cell in the battery pack, including single cell Battery voltage ,temperature , deformation pressure , gas concentration One or more of; S2, clustering the single cells based on the status data, and then dynamically selecting a reference cluster; S3, based on the benchmark clustering, calculating the maximum similarity distance between each cluster and performing threshold comparison, and then outputting the fault warning classification result for each cluster.

2. The battery failure early warning method based on improved cluster analysis according to claim 1, characterized in that: In step S2, the state data associated with each single cell is normalized to obtain multi-dimensional characteristic time series data associated with each single cell. The normalization method is expressed by the following formula (1): （1） In formula (1), Represents the normalized value; Indicates one of the state data; Respectively are the minimum and maximum values ​​of the same state data within the acquisition time window.

3. The battery failure early warning method based on improved cluster analysis according to claim 1 or 2, characterized in that: In step S2, the method for clustering the single cells includes the following steps: A1, calculate the weighted similarity distance between single cells , Indicates single battery and single battery The weighted similarity distance between them; A2, set the scanning radius , and Single battery and single battery Add to the matching set; A3, calculate single battery The weighted similarity distance of each single battery in the matching set is determined to be less than or equal to the scanning radius. , If so, the single cell Add to the matching set; A4. After all the single cells have been judged as above, the remaining single cells are added to the non-compliant set.

4. The battery failure early warning method based on improved cluster analysis according to claim 3, characterized in that: For single battery and single battery Voltage similarity distance between , temperature similarity distance , deformation pressure similarity distance , gas concentration similarity distance The weighted average of one or more of .

5. The battery failure early warning method based on improved cluster analysis according to claim 4, characterized in that: Single battery and single battery The similarity distance between the same state data The calculation method is expressed by the following formula (2): （2） Indicates the minimum path The value of each node; represents the total number of nodes in the minimum path; The minimum path is found by the following method steps: C1, for calculation Construct the corresponding matrix; C2: Find a path with the minimum sum of element values ​​from the upper right corner element to the lower left corner element of the matrix, and use it as the minimum path found.

6. The battery failure early warning method based on improved cluster analysis according to claim 5, characterized in that: Step C1 is to calculate The method of constructing the corresponding matrix is ​​expressed by the following formula (3): （3） Indicates the first Liedi The value of the row's elements; Indicates the number of cells per moment within the acquisition time window. The same type of state data is collected into the first state data time series state data values, wherein the elements in the first state data time series are arranged in the order of collection time; Indicates that the single battery The same acquisition time window for collecting status data, for the same type of status data, for the single battery The first time series of the second state data collected state data values, wherein the elements in the second state data time series are arranged in the order of collection time; Indicates the first Column, No. The element value of the row; Indicates the first Column, No. The element value of the row; Indicates the first Column, No. The element value of the row.

7. The battery failure early warning method based on improved cluster analysis according to claim 6, characterized in that: In step C2, the constraints for finding the minimum path are: The upper right corner element in the matrix is ​​the starting point of the path, the lower left corner element is the end point of the path, and As the current node The next node of , eventually forming the minimum path.

8. The battery failure early warning method based on improved cluster analysis according to claim 1, characterized in that: In step S2, the method for dynamically selecting the benchmark cluster includes the following steps: S21, determining whether the clustering result of each of the single cells is that the number of the non-compliant sets is at least 1 and the number of the compliant sets is 0, If yes, an abnormal alarm is issued for the battery pack; If not, go to step S22; S22, determining whether the number of the matching set is 1 and the number of the non-matching set is 0, If yes, the battery pack is determined to be normal and the fault warning process is terminated; If not, the only matching set or the matching set with the largest number of single cells is selected as the reference cluster, and the remaining matching sets or non-matching sets are each a remaining cluster.

9. The battery failure early warning method based on improved cluster analysis according to claim 1 or 8, characterized in that: Step S3 specifically includes the following steps: S31, calculating the inter-cluster similarity distance between each remaining cluster except the benchmark cluster and the benchmark cluster; S32, setting alarm threshold intervals corresponding to different fault warning levels; S33, judging whether the inter-class similarity distance falls within the corresponding alarm threshold range, If yes, classify the remaining clusters into the fault warning level corresponding to the alarm threshold interval they fall into and issue an alarm; If not, it is determined that each battery cell in the remaining cluster operates normally.

10. The battery failure early warning method based on improved cluster analysis according to claim 9, characterized in that: In step S31, the method for calculating the inter-class similarity distance is: The similarity distance between each single battery in the same remaining cluster and each single battery in the reference cluster is calculated, and the maximum similarity distance calculated is used as the inter-cluster similarity distance between the remaining cluster and the reference cluster.

Images