Battery consistency self-adaptive detection method based on self-iteration robust framework

Abstract

The invention discloses a battery consistency self-adaptive detection method based on a self-iteration robust framework, and the method comprises the steps: collecting an SOC value sequence of a battery pack, carrying out the initial segmentation of the sequence, carrying out the dynamic optimization through combining with an aggregation algorithm, and achieving the self-adaptive SOC interval division; constructing voltage deviation distribution in each interval and generating a consistency covariant index; comparing the consistency covariant index to an adaptive threshold updated based on the deviation feedback self-iteration to trigger a real-time consistency alert; and predicting the consistency level of the next SOC interval through the prediction model by using the temperature information entropy enhancement feature and the historical covariant exponential sequence, thereby realizing prospective early warning. According to the method, the SOC partition can better fit the actual operation state, and the stability of deviation calculation is improved; the robustness of consistency judgment is improved through self-adaptive updating of a threshold value; in combination with real-time detection and trend prediction, the risk can be identified in advance before consistency degradation, and the safety and the pre-judgment capability of the battery management system are significantly enhanced.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to an adaptive detection method for battery consistency based on a self-iterative robust framework. Background Technology

[0002] In the global trend of energy transition and sustainable development, the importance of battery technology is becoming increasingly prominent. Taking new energy vehicles as an example, their power source is highly dependent on the power battery pack, and the consistency between individual cells directly affects the overall performance: since the actual efficiency of the battery pack is limited by the weakest cell, cells with poor consistency are more likely to reach the charge / discharge cutoff voltage earlier, resulting in a significant reduction in the system's usable capacity, thereby limiting the vehicle's range and power output. In addition, this inconsistency will frequently trigger the protection mechanisms of the battery management system (BMS) and the vehicle controller, causing vehicle operation to be restricted; under extreme conditions, it may even cause a sudden interruption of battery power, resulting in reduced braking performance or drive system failure, thus bringing significant safety hazards. This series of problems not only shortens the battery's lifespan but also poses a potential threat to driving safety.

[0003] Similar issues are not limited to power batteries; they are equally prominent in the field of energy storage batteries. Energy storage batteries are widely used in scenarios such as grid peak shaving, photovoltaic / wind power systems, microgrids, or backup power supplies. Inconsistencies can lead to low system efficiency, accelerated capacity degradation, and even thermal runaway or system failure. For example, in large-scale energy storage systems, inconsistencies in individual battery cells can be amplified to the entire battery pack, affecting energy conversion efficiency, increasing maintenance costs, and potentially threatening grid stability. Therefore, developing a universal battery consistency testing method is crucial for improving the overall performance, safety, and lifespan of battery packs.

[0004] While existing methods can quantify battery consistency, they have several problems, including: 1. They rely on preset parameters and cannot adapt to real-time environmental changes; 2. They ignore the model's self-iterative mechanism and cannot adaptively optimize the threshold weights based on new data, resulting in a significant decrease in accuracy as the number of battery cycles increases.

[0005] Therefore, it is necessary to propose an adaptive detection method for battery consistency that can perform adaptive battery SOC partitioning and introduce a self-iterative update mechanism so that the safety threshold can be dynamically adjusted according to deviation feedback and alarm status, thereby achieving more accurate and forward-looking assessment and monitoring of battery consistency. Summary of the Invention

[0006] This invention provides a battery consistency adaptive detection method based on a self-iterative robust framework, addressing technical problems in existing battery consistency detection technologies such as unreasonable segmentation, fixed thresholds, unstable results, and inability to provide early warnings. The technical solution is as follows:

[0007] In a first aspect, embodiments of the present invention provide a battery consistency adaptive detection method based on a self-iterative robust framework, comprising:

[0008] Real-time acquisition of the battery pack's SOC value sequence;

[0009] Perform initial equal-interval segmentation on the SOC value sequence to obtain an initial SOC boundary set;

[0010] The initial boundary of the SOC is dynamically optimized based on the aggregation algorithm to obtain the aggregated optimized boundary set.

[0011] The aggregation optimization boundary set is updated online according to the update cycle to form an adaptive SOC partition;

[0012] Within each SOC partition, the individual cell voltage deviation sequence is calculated, and a consistency covariance index is constructed based on the deviation distribution;

[0013] The consistency covariance index is compared with an adaptive threshold, and a real-time consistency alarm is triggered when it is greater than the threshold; wherein, the adaptive threshold is updated iteratively based on the alarm status and deviation feedback.

[0014] In some embodiments of the present invention, the method further includes:

[0015] Based on the temperature information entropy enhancement feature and the historical covariance index sequence, the consistent covariance index of the next interval is predicted using a prediction model, and the predicted value of the covariance index is obtained.

[0016] The predicted value of the covariance index is compared with the target consistent covariance index. When the predicted value of the covariance index is greater than the target consistent covariance index, a forward consistency alarm is generated.

[0017] Secondly, the present invention also provides a battery consistency adaptive detection device based on a self-iterative robust framework, comprising:

[0018] The data acquisition module is used to collect the SOC value sequence of the battery pack in real time.

[0019] The initial boundary partitioning module is used to perform initial equal-interval segmentation on the SOC value sequence to obtain the initial SOC boundary set;

[0020] The aggregation optimization module is used to dynamically optimize the initial boundary of the SOC based on the aggregation algorithm to obtain the aggregation optimization boundary set;

[0021] The adaptive partitioning module is used to update the aggregated optimization boundary set online according to the update cycle to form an adaptive SOC partition.

[0022] The consistency judgment module is used to calculate the individual cell voltage deviation sequence within each SOC partition and construct a consistency covariance index based on the deviation distribution.

[0023] The output and update module is used to compare the consistency covariance index with the adaptive threshold, and trigger a real-time consistency alarm when it is greater than the threshold; wherein, the adaptive threshold is self-iteratively updated based on the alarm status and deviation feedback.

[0024] Thirdly, the present invention also provides an electronic device, comprising: a processor and a memory;

[0025] The memory stores a computer-readable program that can be executed by the processor;

[0026] When the processor executes the computer-readable program, it implements the steps in the battery consistency adaptive detection method based on a self-iterative robust framework as described above.

[0027] The beneficial effects of the present invention include at least the following:

[0028] (1) The false alarm rate is significantly reduced by using an adaptive SOC partitioning strategy and an adaptive threshold:

[0029] An adaptive SOC partition is constructed by using an "initial segmentation + aggregation optimization + online update" approach, which allows the partition to automatically adjust according to the battery's operating status, thus avoiding the problem that traditional fixed segmentation cannot reflect the voltage characteristics differences between different SOC ranges.

[0030] (2) Information entropy enhancement features enable noise smoothing and real-time tracking, overcoming the shortcomings of existing patents in environmental noise conditions:

[0031] By introducing temperature information entropy and consistency covariance index based on temperature deviation distribution, it is more robust than traditional standard deviation and range, maintaining high sensitivity under noise, temperature fluctuations, and multimodal distributions. This combination significantly improves the accuracy of inconsistency detection and effectively reduces false positives and false negatives.

[0032] (3) Integrating a hierarchical response architecture and predictive maintenance:

[0033] By employing bias feedback and learning rate updates, the threshold can automatically converge and adjust its sensitivity based on battery status, avoiding the high false alarm rate caused by fixed thresholds during the aging phase. Simultaneously, this invention introduces information entropy as an enhancing feature into prediction models such as ARIMA and LSTM to achieve forward prediction of the consistency covariance index for the next cycle, constructing a hierarchical response architecture of "real-time detection + predictive early warning." This mechanism enables the system to identify risks in advance, evolving from passive alarms to predictive maintenance.

[0034] This method is applicable to various types of energy storage batteries and application scenarios, such as grid peak shaving, photovoltaic energy storage systems, or microgrids. It does not rely on fixed thresholds or static models. By integrating real-time data analysis and prediction algorithms, it enables early detection and intervention of abnormal battery voltage, thereby extending battery life and improving system safety. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 This is a system architecture diagram of an embodiment of the battery consistency adaptive detection method based on a self-iterative robust framework provided by the present invention;

[0037] Figure 2 This is a flowchart of an embodiment of the adaptive battery consistency detection method based on a self-iterative robust framework provided by the present invention.

[0038] Figure 3 This is a schematic diagram of an embodiment of the battery consistency adaptive detection device based on a self-iterative robust framework provided by the present invention.

[0039] Figure 4 A schematic diagram of the operating environment of an embodiment of the electronic device provided by the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0041] It should be noted that the battery consistency adaptive detection method based on the self-iterative robust framework provided in this application is generally executed by the terminal device, and correspondingly, the battery consistency adaptive detection device based on the self-iterative robust framework is generally set in the terminal device.

[0042] Figure 1 An exemplary system architecture is shown for a battery consistency adaptive detection method or a battery consistency adaptive detection device based on a self-iterative robust framework that can be applied to this application.

[0043] like Figure 1As shown, the system architecture may include: terminal device 101 and server 102. Terminal device 101 and server 102 can communicate via a network, which serves as the medium for providing communication links between the various units. The network may include various types of wired or wireless communication links, such as: wired communication links including fiber optic cables, twisted-pair cables, or coaxial cables; and wireless communication links including Bluetooth communication links, Wi-Fi communication links, or microwave communication links.

[0044] It should be noted that the terminal device 101 and the server 102 can be either hardware or software. When the terminal device 101 and the server 102 are hardware, they can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When the terminal device 101 and the server 102 are software, they can be implemented as multiple software programs or software modules (for example, to provide distributed services), or as a single software program or software module; no specific limitations are made here.

[0045] The terminal device of this application can be equipped with various communication client applications, such as video recording applications, video playback applications, voice interaction applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0046] A terminal device can be either hardware or software. When the terminal device is hardware, it can be various terminal devices with a display screen, including but not limited to smartphones, tablets, laptops, and desktop computers. When the terminal device is software, it can be installed on the terminal devices listed above. It can be implemented as multiple software programs or software modules (e.g., used to provide distributed services) or as a single software program or software module; no specific limitation is made here.

[0047] When the terminal device is hardware, it can also be equipped with a display device and a camera. The display device can be any device capable of displaying information, and the camera is used to capture video streams. For example, the display device can be a cathode ray tube display (CR), a light-emitting diode display (LED), an e-ink screen, a liquid crystal display (LCD), a plasma display panel (PDP), etc. Users can use the display device on the terminal device to view displayed text, images, videos, and other information.

[0048] It should be understood that Figure 1The number of terminal devices, networks, and servers shown is for illustrative purposes only. Depending on implementation needs, there can be any number of terminal devices, networks, and servers.

[0049] The following will be combined with the appendix Figure 2 This application provides a detailed description of a battery consistency adaptive detection method based on a self-iterative robust framework, as provided in the embodiments of this application. Specifically, the battery consistency adaptive detection device based on a self-iterative robust framework in the embodiments of this application can be... Figure 1 The terminal device shown.

[0050] Please see Figure 2 This document provides a flowchart illustrating an adaptive battery consistency detection method based on a self-iterative robust framework, as described in an embodiment of this application. Figure 2 As shown, the method described in this application embodiment may include the following steps:

[0051] S201. Real-time acquisition of the SOC value sequence of the battery pack;

[0052] S202. Perform initial equal-interval segmentation on the SOC value sequence to obtain the initial SOC boundary set;

[0053] S203. Dynamically optimize the initial boundary of the SOC based on the aggregation algorithm to obtain the aggregated optimized boundary set;

[0054] S204. Update the aggregation optimization boundary set online according to the update cycle to form an adaptive SOC partition;

[0055] S205. Calculate the individual cell voltage deviation sequence within each partition of the adaptive SOC partition, and calculate the consistency covariance index based on the deviation distribution.

[0056] S206. Compare the consistency covariance index with the adaptive threshold, and trigger a real-time consistency alarm when the index is greater than the adaptive threshold; wherein the adaptive threshold is updated iteratively based on the alarm status and deviation feedback.

[0057] The method in this embodiment first acquires the SOC sequence of the battery pack in real time; second, it segments the battery pack into initial equal intervals to construct initial SOC boundaries; third, it dynamically optimizes the boundaries based on an aggregation algorithm; and finally, it updates the boundaries online according to a preset period to form adaptive SOC partitions, thereby enabling the SOC range to automatically adjust with operating conditions, temperature, and battery aging. Based on this, the method calculates the voltage deviation of individual cells within each adaptive SOC partition and constructs a consistency covariance index. Real-time consistency is determined using an adaptive threshold, triggering an alarm when the threshold is exceeded. This threshold is continuously iteratively updated based on historical deviation feedback and alarm results. Through the closed-loop operation of adaptive partitioning, the covariance index, and the self-learning threshold, the method in this embodiment enables the monitoring process to dynamically match changes in battery characteristics, improving the ability to identify minor degradation and inconsistency trends.

[0058] In this embodiment, for step S201, a state of charge (SOC) range with relatively stable voltage characteristics is selected, and the SOC value sequence of the battery pack is collected in real time in the battery management system (BMS). For example, in 500 data points, the SOC value ranges from 0% to 100%, the sampling frequency is once per second, and the total collection time is about 8 minutes.

[0059] After data acquisition, the data is preprocessed, including noise removal. For example, the SOC sequence is smoothed using the average of 5 data points: if the original SOC value is [3%, 4%, 5%, 6%, 7%], the smoothed value is 5%. Simultaneously, the voltage values ​​of 96 individual cells (typical electric vehicle battery pack configuration) are recorded, with an average voltage of approximately 3.7V.

[0060] In this embodiment, the SOC value sequence is initially coarsely segmented in step S202:

[0061] For the collected SOC sequences Perform equal-interval D% segmentation to generate an initial boundary set, expressed by the formula:

[0062]

[0063] in, For the initial boundary set, This represents the i-th boundary point.

[0064] For example, the collected SOC sequence is divided into segments with equal intervals of 5% to generate an initial set of boundaries, such as boundary points of 0%, 5%, 10%, 15%, 20%, ..., 95%, 100% (a total of 21 boundary points, generating 20 initial intervals).

[0065] Furthermore, to ensure sufficient samples in each initial interval and avoid fragmentation caused by starting aggregation from scratch, the number of samples in each SOC interval is not less than a preset minimum sample threshold, improving the robustness of subsequent aggregation. That is, the number of samples in each SOC interval satisfies:

[0066] ≥ ,

[0067] in, Let be the number of samples in the k-th interval. The minimum sample threshold is used in practice. The value is typically set to 20. Each interval should have at least 20 data points to avoid instability in subsequent processing. If an interval has fewer than 20 samples (e.g., only 15), it should be merged into an adjacent interval to maintain robustness.

[0068] In this embodiment, for step S203, based on the initial segmentation, the collected SOC sequence data is processed... ( Given a SOC data sequence (t=1 to N, where N is the total number of data points), a clustering algorithm is applied as an adaptive aggregation algorithm to output M cluster centers. The initial SOC interval boundaries are then adjusted as follows:

[0069]

[0070] in, For the cluster center set, Denotes the j-th cluster center. This is the boundary set after aggregation optimization.

[0071] It should be noted that the aggregation algorithm is one of K-means, GMM, or weighted clustering algorithm;

[0072] The initial boundary of the SOC is dynamically optimized based on the aggregation algorithm to obtain the set of aggregated optimized boundaries, including: obtaining the cluster center by minimizing the weighted Euclidean distance, and using the midpoint of the cluster center as the aggregated optimized boundary.

[0073] Minimize the variance within a multimodal weighted cluster Represented as:

[0074]

[0075] in, For weighted Euclidean distance, the weights (set up To highlight the core variable SOC). Let k be the set of data points for the k-th cluster. It is the center of the k-th cluster.

[0076] The smaller the variance J, the closer the data of each cluster is to its center, indicating that each SOC interval is more uniform and that the segmentation method is more reasonable. Therefore, by minimizing J, we can find the optimal SOC segmentation method with the smallest internal difference. Through cluster optimization, we can make the SOC segmentation structure match the actual SOC data distribution and voltage change pattern, thereby automatically generating a more reasonable SOC partition and improving the stability and accuracy of subsequent consistency covariance index calculation.

[0077] As a specific implementation, the K-means algorithm is used as an adaptive aggregation algorithm to divide 500 SOC data points into 5 clusters. The weighted distance from each SOC point to the cluster center is calculated (the weight is set to 0.8, prioritizing SOC values).

[0078] Output example: 5 cluster centers with values ​​of 12%, 38%, 62%, 78%, and 95%.

[0079] That is: find the natural center points of the 5 cluster centers using K-means, and use the midpoint between two centers as the segment boundary. Adjust the boundary to the midpoint of the cluster center. The new boundary is 0%, (12%+38%) / 2, (38%+62%) / 2, (62%+78%) / 2, (78%+95%) / 2, 100%, which is 0%, 25%, 50%, 70%, 86.5%, and 100%.

[0080] In this embodiment, step S204, which involves updating the boundary set online according to the update cycle, includes:

[0081] The old boundary points and the latest cluster centers of the current period are weighted and updated according to the decay factor to achieve periodic convergence of the SOC interval boundary.

[0082] Specifically, based on the current period's aggregated optimization As initial ( (This is the set of old boundaries for the current period), and an update is performed every period P (P is a preset update period, for example, P=100 data points) to obtain each internal boundary point. New boundary points after update This can be expressed as a formula:

[0083]

[0084] Where β is the decay factor (e.g., 0.1). As the center of the newly aggregated j clusters, Let J be the j-th old boundary point, j=2 to M.

[0085] Then the i-th SOC interval is:

[0086]

[0087] After the update, check each interval. If the new boundary leads to uneven sampling If this is the case, then adjacent clusters will be forcibly merged to maintain robustness.

[0088] Furthermore, the updated boundary set is obtained based on all the updated boundary points. This is used for calculating the consistent covariance index for the next period.

[0089] As a specific embodiment, the method for performing online updates and boundary adjustments in step S204 is as follows:

[0090] Starting from the aggregated and optimized boundary, updates are performed every 100 data points (approximately 1.7 minutes). For each internal boundary point, for example, if the old boundary is 25%, and the latest cluster center calculates a new adjustment value of 28%, then the new boundary is updated with a decay factor of 0.1 to obtain the new boundary = 25% + 0.1 × (28% - 25%) = 25.3%.

[0091] After updating, check each interval. If the number of samples is less than 20 (e.g., an interval has only 18 samples), merge adjacent intervals. For example, merge 25.3%-50% and 50%-70% into 25.3%-70%.

[0092] The final output set of new boundaries, such as 0%, 22%, 45%, 68%, 82%, and 100%, is used for the calculation of the consistent covariance exponent in the next cycle. The i-th SOC interval is defined as the interval from one boundary to the next, for example, the first interval is 0%-22%.

[0093] In some embodiments, after segmenting the SOC interval in step S205, the voltage value sequence of the N individual cells in the battery pack is collected within the i-th SOC interval:

[0094]

[0095] Calculate the voltage deviation sequence

[0096]

[0097] in, , The average voltage. .

[0098] The voltage deviation sequence is discretized into K intervals according to a preset number of bins K. The occurrence frequency of deviation points in each interval is counted and the probability distribution is calculated. .

[0099] Based on the aforementioned probability distribution, the uniform covariance index, calculated from the voltage deviation distribution, can be obtained, expressed by the following formula:

[0100]

[0101] in, Let be the probability distribution of voltage deviation within the discretized interval t.

[0102] In step S206, the calculation of the consistent covariance index employs a sliding window mechanism. The probability distribution and statistics are calculated within a time window of length L, and the window is updated with a step size s to obtain a time-varying consistent covariance index sequence. The consistent covariance exponent is combined with the optimized threshold. In comparison, if > If this occurs, an alarm will be triggered for the current SOC range.

[0103] To achieve harmonic threshold optimization, this embodiment introduces a dynamic update mechanism, specifically:

[0104] Initial threshold Based on historical data or experience (e.g.) =0.5); after every t-th SOC, if there is no alarm, calculate the deviation feedback. and update the threshold , where α is the learning rate (e.g., 0.01), which allows the threshold to adaptively adjust according to the change in the consistency covariance exponent, enhancing the sensitivity to transitions in the SOC range.

[0105] As a specific implementation, after completing step S204, the consistency covariance index is calculated and determined under dynamic segmentation, including:

[0106] For each dynamic segment interval (e.g., 0%-25.3%, 25.3%-50.2%, 50.2%-70%, 70%-86.5%, 86.5%-100%), calculate the voltage deviation of 96 individual cells. For example, in the first interval, the average voltage is 3.65V, and the deviation of a single cell is 0.03V, 0.05V, etc. (total deviation range 0-0.2V).

[0107] Then, the uniform covariance index is calculated based on the voltage deviation distribution. ,in This represents the probability distribution of voltage deviation within the discretized interval. The deviation is divided into 8 groups (each group width is 0.05V, such as 0-0.05V, 0.05V-0.1V, etc.), and the probability of each group is calculated, for example, the probability is [0.15, 0.20, 0.18, 0.12, 0.10, 0.08, 0.09, 0.08].

[0108] Then the uniform covariance index ≈ 2.85 (out of 3, normalized range 0-1).

[0109] The consistency slope index and the adaptive threshold In comparison, if > If this occurs, an alarm will be triggered for the current SOC range. For example, in an aging battery, the alarm will be triggered 3 times.

[0110] Furthermore, to achieve harmonic threshold optimization, some embodiments of the present invention also introduce a dynamic update mechanism for the consistency covariance index, including:

[0111] Initial threshold Based on historical data or experience (e.g.) =0.5); after each t-th SOC interval (e.g., t=5), if there is no alarm, let... , Calculate deviation feedback (equal to the current) Subtract the previous one (Approximately equal to 0.03), and update the threshold. Where α is the learning rate (e.g., 0.01), this allows the threshold to adaptively adjust according to the change in the consistency covariance exponent, enhancing sensitivity to transitions within the SOC range. Example: Initial T=0.5, D=0.03, then... After 10 updates, It can be adjusted from 0.5 to 0.52.

[0112] Current judgment: If H=0.48 < T=0.52, it indicates health and there is no real-time alarm; if H=0.95 > T=0.52, an alarm will be triggered.

[0113] In some embodiments, the method further includes: consistent covariance index prediction and forward alert based on statistical or machine learning models.

[0114] Current Consistent Covariance Index When there is no real-time alarm when the value is below the threshold, to improve the foresight of battery consistency monitoring, we use statistical or machine learning models to predict the consistency covariance exponent for the next SOC interval (t+1). This is based on the current information entropy of the temperature deviation distribution. Integrating this as an enhancement feature into the model enables the input to capture the uncertainty of consistent computation under temperature fluctuations (such as high entropy). A value > 0.9 indicates a potential risk of decay, which enhances sensitivity, especially during transitions within the SOC range.

[0115] in:

[0116]

[0117] This method forms a closed-loop prediction mechanism, which, combined with dynamic segmentation and adaptive threshold adjustment, improves robustness compared to traditional real-time calculation methods, reducing alarm response time to 1.5 minutes. Specifically:

[0118] Assume a dynamic range (e.g., 0%-25.3%, 25.3%-50.2%), and collect data from the past 20 ranges. For a sequence (e.g., [0.92, 0.89, ..., 0.95]), calculate the difference. ≈ 0.03 ≈0.2, ≈3.65V, and with [ , , [] as input.

[0119] Model options include ARIMA (p=2, d=1, q=1, m=5) or LSTM (64 hidden units), trained on 500 historical sequences (MSE=0.015). Example prediction: Input [0.95, 0.92, 0.03], Output .like This triggers forward alerts, such as outputting an alert that "the risk in the next interval has increased."

[0120] After the time rolls to t+1, a new threshold is generated. (α=0.01, ≈0.02), calculate the actual (e.g., 0.79), if If the value is 0.52, a real-time alarm will be triggered, and then the prediction will be made for t+2.

[0121] By introducing a trainable predictive model and a self-iterative threshold update mechanism, risks can be identified in advance before consistency deterioration occurs, while the judgment criteria are automatically calibrated when actual consistency fluctuations occur. This mechanism significantly reduces false alarms and false negatives, making consistency detection both forward-looking and robust in real time, thereby improving the safety and stability of battery operation.

[0122] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0123] Please see Figure 3 This illustration shows a schematic diagram of a battery consistency adaptive detection device based on a self-iterative robust framework, hereinafter referred to as device 3, provided in an exemplary embodiment of this application. Device 3 can be implemented as all or part of a terminal device through software, hardware, or a combination of both. Device 3 includes:

[0124] Data acquisition module 310 is used to acquire the SOC value sequence of the battery pack in real time;

[0125] The initial boundary partitioning module 320 is used to perform initial equal-interval segmentation on the SOC value sequence to obtain an initial SOC boundary set.

[0126] The aggregation optimization module 330 is used to dynamically optimize the initial boundary of the SOC based on the aggregation algorithm to obtain the aggregation optimization boundary set.

[0127] The adaptive partitioning module 340 is used to update the aggregated optimization boundary set online according to the update cycle to form an adaptive SOC partition;

[0128] The consistency judgment module 350 is used to calculate the single cell voltage deviation sequence within each SOC partition and construct a consistency covariance index based on the deviation distribution.

[0129] The output and update module 360 ​​is used to compare the consistency covariance index with the adaptive threshold, and trigger a real-time consistency alarm when it is greater than the threshold; wherein, the adaptive threshold is self-iteratively updated based on the alarm status and deviation feedback.

[0130] Please see Figure 4 This document provides a schematic diagram of the structure of a terminal device according to an embodiment of this application. Figure 4 As shown, the terminal device 400 may include: at least one processor 401, at least one network interface 404, user interface 403, memory 405, and at least one communication bus 402.

[0131] The communication bus 402 is used to enable communication between these components.

[0132] The user interface 403 may include a display screen and a camera. Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0133] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0134] The processor 401 may include one or more processing cores. The processor 401 connects to various parts within the terminal device 400 using various interfaces and lines, and performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 405, and by calling data stored in the memory 405. Optionally, the processor 401 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 401 may integrate one or more combinations of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip, without being integrated into the processor 401.

[0135] The memory 405 may include random access memory (RAM) or read-only memory. Optionally, the memory 405 may include a non-transitory computer-readable storage medium. The memory 405 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 405 may also be at least one storage device located remotely from the aforementioned processor 401. Figure 4 As shown, the memory 405, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and application programs.

[0136] exist Figure 4 In the terminal device 400 shown, the user interface 403 is mainly used to provide an input interface for the user and to obtain the user's input data; while the processor 401 can be used to call the application program stored in the memory 405 and specifically execute, such as Figure 2 The method shown can be referred to for details. Figure 2 As shown, it will not be elaborated further here.

[0137] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory, or random access memory, etc.

[0138] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A battery consistency adaptive detection method based on a self-iterative robust framework, characterized in that, include: Real-time acquisition of the battery pack's SOC value sequence; Perform initial equal-interval segmentation on the SOC value sequence to obtain an initial SOC boundary set; The initial boundary of the SOC is dynamically optimized based on the aggregation algorithm to obtain the aggregated optimized boundary set. The aggregation optimization boundary set is updated online according to the update cycle to form an adaptive SOC partition; Within each SOC partition, the individual cell voltage deviation sequence is calculated, and a consistency covariance index is constructed based on the deviation distribution. The consistency covariance index is compared with an adaptive threshold, and a real-time consistency alarm is triggered when it is greater than the threshold; wherein, the adaptive threshold is updated iteratively based on the alarm status and deviation feedback.

2. The method according to claim 1, characterized in that, The method further includes: Based on the temperature information entropy enhancement feature and the historical covariance index sequence, the consistent covariance index of the next interval is predicted using a prediction model, and the predicted value of the covariance index is obtained. The predicted value of the covariance index is compared with the target consistent covariance index. When the predicted value of the covariance index is greater than the target consistent covariance index, a forward consistency alarm is generated.

3. The method according to claim 1, characterized in that, The initial equal-interval segmentation of the SOC value sequence includes: The SOC range is divided into multiple initial SOC intervals at equal intervals from 0% to 100%, and the dividing point between adjacent intervals is used as the initial SOC boundary. The number of samples in each SOC interval is not less than the preset minimum sample threshold.

4. The method according to claim 1, characterized in that, The aggregation algorithm is one of K-means, GMM, or weighted clustering algorithm; The initial boundary of the SOC is dynamically optimized based on the aggregation algorithm to obtain the aggregation optimization boundary set, including: obtaining the cluster center by minimizing the weighted Euclidean distance, and using the midpoint of the cluster center as the aggregation optimization boundary.

5. The method according to claim 1, characterized in that, The online update of the boundary set according to the update cycle includes: The old boundary points and the latest cluster centers of the current period are weighted and updated according to the decay factor to achieve periodic convergence of the SOC interval boundary.

6. The method according to claim 1, characterized in that, The method for calculating the uniform covariance index includes: The voltage deviation is discretized into multiple sub-intervals with a preset width, and the probability distribution of the deviation in each interval is calculated. The temperature information entropy is calculated based on the probability distribution, and then normalized to obtain the consistent covariance index.

7. The method according to claim 1, characterized in that, The method for self-iteratively updating the adaptive threshold based on alarm status and deviation feedback includes: Based on the feedback of the deviation between the consistent covariance index of the previous cycle and the current consistent covariance index, the adaptive threshold is iteratively adjusted according to the preset learning rate. When a consistency alert occurs, increase the update magnitude of the threshold to improve sensitivity to anomalies; When no consistent alarm occurs, the update magnitude of the threshold is reduced to decrease the false alarm rate.

8. The method according to claim 2, characterized in that, The prediction model is one of ARIMA, LSTM, GRU, or a sequence prediction model based on Bayesian optimization. The prediction model uses the historical consistent covariance index sequence, temperature information entropy enhancement feature, SOC sequence, and voltage deviation statistics as input features to predict the consistent covariance index for the next SOC interval, thus obtaining the predicted covariance index value.

9. A battery consistency adaptive detection device based on a self-iterative robust framework, characterized in that, include: The data acquisition module is used to collect the SOC value sequence of the battery pack in real time. The initial boundary partitioning module is used to perform initial equal-interval segmentation on the SOC value sequence to obtain the initial SOC boundary set; The aggregation optimization module is used to dynamically optimize the initial boundary of the SOC based on the aggregation algorithm to obtain the aggregation optimization boundary set; The adaptive partitioning module is used to update the aggregated optimization boundary set online according to the update cycle to form an adaptive SOC partition. The consistency judgment module is used to calculate the individual cell voltage deviation sequence within each SOC partition and construct a consistency covariance index based on the deviation distribution. The output and update module is used to compare the consistency covariance index with the adaptive threshold, and trigger a real-time consistency alarm when it is greater than the threshold; wherein, the adaptive threshold is self-iteratively updated based on the alarm status and deviation feedback.

10. An electronic device, characterized in that, include: Processor and memory; The memory stores a computer-readable program that can be executed by the processor; When the processor executes the computer-readable program, it implements the steps in the battery consistency adaptive detection method based on a self-iterative robust framework as described in any one of claims 1-8.

Images