Battery system fault detection methods, apparatus, and non-transitory storage media

Abstract

The application discloses a battery system fault detection method, device and nonvolatile storage medium. The method comprises the following steps: obtaining multi-level operation data of a target battery system at multiple preset sampling time points within a preset collection period; extracting a first feature, a second feature and a third feature based on the multi-level operation data, and splicing the first feature, the second feature and the third feature into a multi-dimensional feature vector; determining a first fault detection result of the target battery system based on the multi-level operation data; determining a second fault detection result of the target battery system based on the multi-dimensional feature vector and the multi-level operation data; and determining a target fault detection result of the target battery system based on the first fault detection result and the second fault detection result. The application solves the problems of single technical data utilization, rigid threshold determination and poor interpretability, realizes early and accurate early warning of energy storage battery faults, and improves the accuracy and interpretability of early warning.

Description

Technical Field

[0001] This invention relates to the field of safety technology for electrochemical energy storage systems, and more specifically, to a method, apparatus, and non-volatile storage medium for detecting battery system faults. Background Technology

[0002] As the core component of electrochemical energy storage systems, energy storage batteries are subjected to the combined effects of complex operating conditions (such as frequent charge-discharge switching and temperature fluctuations) and aging processes over long periods, making them prone to faults such as thermal runaway, overcharging, over-discharging, and internal short circuits. Current early warning technologies are mostly based on single data sources (such as only individual cell voltage or temperature), which makes it difficult to comprehensively reflect the true state across levels (individual cells, battery packs, and battery clusters). Furthermore, they lack cross-scale correlation analysis of early signals propagating from individual cells to the system, failing to reflect the spatiotemporal continuity of fault evolution. In addition, traditional fault diagnosis often focuses on static threshold judgment or shallow data-driven models, failing to deeply integrate electrochemical mechanism knowledge with deep learning features. This results in low sensitivity to early, minor fault identification, and the output results are often limited to qualitative judgments, lacking interpretable analysis of key influencing factors and fine-grained fault situation awareness, making it difficult to meet the needs of precise operation and maintenance and early warning.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a battery system fault detection method, apparatus, and non-volatile storage medium to at least solve the technical problem that traditional methods, which rely solely on individual data, cannot capture early signals of fault propagation from individual cells to battery packs and battery clusters due to the single and isolated data source, lack of cross-scale correlation analysis, and reliance on individual cell data.

[0005] According to one aspect of the present invention, a battery system fault detection method is provided, comprising: acquiring multi-level operational data of a target battery system at multiple preset sampling time points within a preset acquisition period, wherein the multi-level operational data includes operational data of multiple individual battery cells, operational data of multiple battery packs, and operational data of multiple battery clusters, wherein multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster; extracting a first feature, a second feature, and a third feature based on the multi-level operational data to form a multi-dimensional feature vector, wherein the first feature characterizes the transient change characteristics of the operational data of individual battery cells, the second feature characterizes the steady-state distribution characteristics of the operational data of individual battery cells, and the third feature characterizes the correlation characteristics between individual battery cells, battery packs, and battery clusters; Based on multi-level operational data, a first fault detection result for the target battery system is determined, whereby the first fault detection result includes whether the target battery system is in normal, warning, or alarm state. Based on multi-dimensional feature vectors and multi-level operational data, a second fault detection result for the target battery system is determined, whereby the second fault detection result includes binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification results characterize whether a fault exists in the target battery system, the continuous value prediction results characterize the severity index of the fault, the trend prediction results characterize the changing trend of the target battery system's operational data, and the list of key influencing factors characterizes the cause of the fault in the target battery system. Based on the first and second fault detection results, a target fault detection result for the target battery system is determined.

[0006] Optionally, within a preset acquisition period, multi-level operational data of the target battery system at multiple preset sampling time points are acquired, including: the voltage of each individual battery cell, the temperature of each individual battery cell, the state of charge of each individual battery cell, the health status of each individual battery cell, the total current of each battery pack, the extreme values ​​of the individual cell voltages of each battery pack, the maximum temperature difference of each battery pack, the total voltage of each battery cluster, the total current of each battery cluster, and the ambient temperature of each battery cluster.

[0007] Optionally, based on multi-level operational data, a first feature, a second feature, and a third feature are extracted to form a multi-dimensional feature vector, including: constructing the first feature by calculating the voltage change rate and temperature change rate of each of the multiple battery cells within a preset acquisition period based on their individual voltages and temperatures; constructing the second feature by calculating the mean voltage, voltage standard deviation, mean temperature, and temperature standard deviation of each of the multiple battery cells within a preset acquisition period based on their individual voltages and temperatures; calculating the maximum single-cell voltage difference of each of the multiple battery packs based on their individual cell voltage extremes; calculating the average voltage standard deviation of all battery packs in the same battery cluster; calculating the voltage-temperature correlation coefficient of each of the multiple battery cells; and constructing the third feature based on the maximum single-cell voltage difference of each of the multiple battery packs, the average voltage standard deviation of the battery pack, and the voltage-temperature correlation coefficient of each of the multiple battery cells.

[0008] Optionally, based on multi-level operational data, the first fault detection result of the target battery system is determined, including: obtaining the historical standard voltage of the individual cells in the target battery system; determining a first coefficient based on the state of charge of each individual cell; determining a second coefficient based on the temperature of each individual cell; determining a third coefficient based on the total current of each battery pack; determining a fourth coefficient based on the health status of each individual cell; calculating the product of the historical standard voltage, the first coefficient, the second coefficient, the third coefficient, and the fourth coefficient to obtain a voltage difference threshold; and comparing the voltage difference threshold with the maximum single-cell voltage difference of each battery pack to obtain the first fault detection result.

[0009] Optionally, the second fault detection result of the target battery system is determined based on the multidimensional feature vector, including: making a preliminary prediction of the fault probability and operating data of the target battery system based on the multidimensional feature vector to form a time-series feature vector; calculating the physical mechanism characteristics of the battery cells in the target battery system based on multi-level operating data to form a physical feature vector; and fusing the time-series feature vector and the physical feature vector and inputting them into a preset gradient boosting decision tree ensemble learning model to obtain the second fault detection result.

[0010] Optionally, based on the first fault detection result and the second fault detection result, the target fault detection result of the target battery system is determined, including: if the first fault detection result is a warning, determining whether the binary classification results of the first fault detection result and the second fault detection result match; if the binary classification results of the first fault detection result and the second fault detection result match, integrating the first fault detection result and the second fault detection result to obtain the target fault detection result.

[0011] According to another aspect of the present invention, a battery system fault detection device is also provided, comprising: an acquisition module, configured to acquire multi-level operating data of a target battery system at multiple preset sampling time points within a preset acquisition period, wherein the multi-level operating data includes operating data of multiple individual battery cells, operating data of multiple battery packs, and operating data of multiple battery clusters, wherein multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster; an extraction module, configured to extract a first feature, a second feature, and a third feature based on the multi-level operating data, and concatenate them into a multi-dimensional feature vector, wherein the first feature characterizes the transient change characteristics of the operating data of individual battery cells, the second feature characterizes the steady-state distribution characteristics of the operating data of individual battery cells, and the third feature characterizes the correlation characteristics between individual battery cells, battery packs, and battery clusters; and a first determination module. The system has three main components: a first determination module and a second determination module. The first determination module determines the first fault detection result of the target battery system based on multi-level operational data. The first fault detection result includes whether the target battery system is in a normal, warning, or alarm state. The second determination module determines the second fault detection result of the target battery system based on multi-dimensional feature vectors and multi-level operational data. The second fault detection result includes binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification results indicate whether the target battery system has a fault, the continuous value prediction results indicate the severity of the fault, the trend prediction results indicate the changing trend of the target battery system's operational data, and the list of key influencing factors indicates the cause of the fault in the target battery system. The third determination module determines the target fault detection result of the target battery system based on the first and second fault detection results.

[0012] According to another aspect of the present invention, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is running, it controls the device where the non-volatile storage medium is located to execute any of the above-described battery system fault detection methods.

[0013] According to another aspect of the present invention, a computer device is also provided, the computer device including a processor, the processor being configured to run a program, wherein the program executes any of the above-described battery system fault detection methods during runtime.

[0014] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements any of the above-described battery system fault detection methods.

[0015] In this embodiment of the invention, a battery system fault detection method is employed. This method acquires multi-level operational data of the target battery system at multiple preset sampling time points within a preset acquisition period. The multi-level operational data includes operational data for each individual battery cell, each battery pack, and each battery cluster. Multiple battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster. Based on the multi-level operational data, a first feature, a second feature, and a third feature are extracted to form a multi-dimensional feature vector. The first feature represents the transient change characteristics of the operational data of a battery cell, the second feature represents the steady-state distribution characteristics of the operational data of a battery cell, and the third feature represents the correlation characteristics between battery cells, battery packs, and battery clusters. Based on the multi-level operational data, a first fault detection result for the target battery system is determined, including whether the target battery system is in a normal, warning, or alarm state. Based on the multi-dimensional feature vector and the multi-level operational data, a second fault detection result for the target battery system is determined. The test results include the following: the second fault detection result includes binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification result indicates whether a fault exists in the target battery system; the continuous value prediction result indicates the severity of the fault; the trend prediction result indicates the changing trend of the target battery system's operating data; and the list of key influencing factors indicates the cause of the fault in the target battery system. Based on the first and second fault detection results, the target fault detection result of the target battery system is determined. This achieves the goal of triggering early warning through a dynamic threshold channel and using a hybrid driving model for dual confirmation and fine-grained information supplementation, realizing the collaborative decision-making of main channel triggering and auxiliary channel verification. This achieves the technical effects of early and accurate early warning, graded diagnosis, and high interpretability of results for energy storage battery faults. Furthermore, it solves the technical problem that traditional methods, which rely solely on individual data and lack cross-scale correlation analysis due to the single and isolated data source, cannot capture early signals of fault propagation from individual cells to battery packs and battery clusters. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0017] Figure 1 A hardware structure block diagram of a computer terminal for implementing a battery system fault detection method is shown.

[0018] Figure 2 This is a schematic flowchart of a battery system fault detection method provided according to an embodiment of the present invention;

[0019] Figure 3This is a flowchart illustrating a multi-level data fusion architecture provided by an optional embodiment of the present invention;

[0020] Figure 4 This is a flowchart illustrating the dynamic threshold adaptive adjustment mechanism provided by an optional embodiment of the present invention;

[0021] Figure 5 This is a schematic diagram of the integrated learning and physics hybrid driving module provided in an optional embodiment of the present invention;

[0022] Figure 6 This is a general flowchart of a method for early warning of energy storage battery faults based on multi-source data fusion and dynamic thresholds according to an optional embodiment of the present invention;

[0023] Figure 7 This is a structural block diagram of a battery system fault detection device provided according to an embodiment of the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] According to an embodiment of the present invention, a battery system fault detection method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0027] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing a battery system fault detection method is shown. Figure 1 As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0028] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0029] The memory 104 can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the battery system fault detection method in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby implementing the aforementioned application program for the battery system fault detection method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0030] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0031] Figure 2 This is a flowchart illustrating a battery system fault detection method provided according to an embodiment of the present invention, such as... Figure 2 As shown, the method includes the following steps:

[0032] Step S201: Within a preset acquisition period, acquire multi-level operating data of the target battery system at multiple preset sampling time points. The multi-level operating data includes the operating data of multiple individual battery cells, the operating data of multiple battery packs, and the operating data of multiple battery clusters. Multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster.

[0033] In this step, the operating data for individual cells includes cell voltage, cell temperature, and state of charge (SOC), used to characterize the electrochemical state inside the cell. The first derivative (rate of change) of the cell voltage sequence is used to reflect the transient characteristics of faults, while the mean and standard deviation within a preset sliding window are used to characterize the steady-state distribution characteristics of the cell state. The operating data for the battery pack includes the total voltage, total current, highest / lowest cell voltage, and maximum temperature difference of the battery pack. Among these, the maximum cell voltage difference among all cells within the same battery pack is extracted as a key feature characterizing the battery pack-level consistency, while the maximum temperature difference is used to characterize the non-uniformity of heat distribution. The operating data for the battery cluster can include the total voltage, total current, and ambient temperature of the battery cluster. By calculating the average voltage standard deviation of all battery packs within the same battery cluster, a feature characterizing the battery cluster-level consistency is constructed; simultaneously, the Pearson correlation coefficient between the voltage and temperature sequences is used to construct a feature characterizing the thermoelectric coupling characteristics. By integrating data from the three levels of individual cells, battery packs, and battery clusters, and extracting time-series features, statistical features, and cross-level correlation features, a high-dimensional multi-dimensional feature vector is formed. This cross-level data fusion architecture overcomes the problem of insufficient global perception capabilities caused by existing technologies relying solely on single-cell data. It can capture early, weak signals of fault propagation from individual cells to the system, providing a comprehensive and multi-dimensional input foundation for subsequent dynamic threshold determination and hybrid-driven prediction models.

[0034] In addition, the collected data needs to undergo various preprocessing steps, such as... Figure 3 This is a flowchart illustrating a multi-level data fusion architecture provided by an optional embodiment of the present invention, such as... Figure 3As shown, the sources of multi-source data can include individual cell voltage, individual cell temperature, total battery pack voltage, total battery pack current, total battery cluster voltage, and ambient temperature. After collection, each type of data undergoes time synchronization, outlier removal, missing value imputation (e.g., linear interpolation), and maximum-minimum normalization to ensure consistency and effectiveness in subsequent modeling. The data sampling interval is set to 30 seconds per data entry, covering various operating conditions such as normal battery charging and discharging, overcharging, over-discharging, and thermal runaway, ensuring the richness and representativeness of the data samples.

[0035] Step S202: Based on multi-level operating data, extract the first feature, the second feature, and the third feature, and concatenate them into a multi-dimensional feature vector. The first feature represents the transient change characteristics of the operating data of a battery cell, the second feature represents the steady-state distribution characteristics of the operating data of a battery cell, and the third feature represents the correlation characteristics between battery cells, battery packs, and battery clusters.

[0036] In this step, the vector is formed by concatenating and fusing features from three dimensions: the first feature focuses on the electrochemical transient response, calculating the rate of change based on the individual cell voltage and temperature sequences, directly reflecting the rapid dynamic characteristics of the battery during charge-discharge switching or the initial stage of a fault; the second feature focuses on the steady-state distribution characteristics, characterizing the voltage balance and temperature stability of the battery under specific operating conditions by calculating the mean and standard deviation within a sliding window, revealing the long-term health status of the battery; the third feature captures cross-level consistency, calculating the maximum voltage difference between individual cells, the standard deviation of voltage between battery packs, and the voltage-temperature correlation coefficient, quantifying the inconsistency and thermoelectric coupling degree from the individual cell to the battery cluster scale, reflecting the early signals of fault propagation from local to system. These three features correspond to transient dynamics, steady-state distribution, and cross-level correlation, respectively. By fusing them into a unified multi-dimensional feature vector, the limitations of traditional single monitoring quantities lacking a global perspective can be overcome. This provides high-dimensional, complementary, and physically meaningful input data for subsequent dynamic threshold determination and hybrid-driven prediction models, thereby significantly improving the ability to capture early weak fault signals and the accuracy of fault diagnosis.

[0037] Step S203: Based on multi-level operating data, determine the first fault detection result of the target battery system, wherein the first fault detection result includes whether the target battery system is in normal, warning, or alarm state.

[0038] In this step, Figure 4 This is a flowchart illustrating the dynamic threshold adaptive adjustment mechanism provided by an optional embodiment of the present invention, as shown below. Figure 4As shown, the current operating state of the battery is first identified based on the magnitude and direction of the real-time current. If the current is greater than 0, it is determined to be in a discharging state; if the current is less than 0, it is determined to be in a charging state; and if the current is equal to 0, it is determined to be in a static state. Subsequently, an adaptive threshold adjustment function is constructed to dynamically calculate the fault judgment threshold. This function adjusts the basic threshold in a multi-dimensional coupling by introducing a state of charge correction coefficient, a temperature correction coefficient, an operating condition correction coefficient, and a health correction coefficient. Specifically, the state of charge correction coefficient takes a value greater than 1 when the SOC is in the extreme range (e.g., <20% or >95%) to relax the threshold, and takes a baseline value in the normal range; the temperature correction coefficient takes a value greater than 1 when the temperature exceeds the safe range (e.g., <10℃ or >45℃) to reflect the accelerated aging effect under high / low temperature conditions; the operating condition correction coefficient takes a value greater than 1 during charging and discharging to consider dynamic stress, and takes a baseline value when static; the health correction coefficient is fine-tuned according to the battery's state of health (SHO). After obtaining the dynamically adjusted threshold, the real-time monitored battery characteristic values ​​(such as maximum single-cell pressure difference or temperature change rate) are compared with the dynamic threshold. Based on the degree of deviation and duration, the battery status is divided into three levels: when the real-time value is less than the warning threshold, it is determined to be in a normal state; when the real-time value is greater than or equal to the warning threshold and continuously exceeds the threshold for a first preset time (e.g., 3 seconds), a warning state is triggered; when the real-time value is greater than or equal to the fault threshold and continuously exceeds the threshold for a second preset time (e.g., 3 seconds), an alarm state is triggered. This dynamic threshold mechanism based on operating status and environmental parameters overcomes the problems of insufficient sensitivity or high false alarm rate of traditional static thresholds under complex operating conditions, achieving accurate graded early warning of energy storage battery faults.

[0039] Step S204: Based on multi-dimensional feature vectors and multi-level operating data, determine the second fault detection result of the target battery system. The second fault detection result includes binary classification result, continuous value prediction result, trend prediction result, and a list of key influencing factors. The binary classification result indicates whether the target battery system has a fault, the continuous value prediction result indicates the severity index of the fault, the trend prediction result indicates the changing trend of the operating data of the target battery system, and the list of key influencing factors indicates the cause of the fault in the target battery system.

[0040] In this step, Figure 5 This is a schematic diagram of the structure of the integrated learning and physics hybrid driving module provided in an optional embodiment of the present invention, as shown below. Figure 5As shown, the hybrid-driven prediction framework includes a data-driven module and a knowledge-guided module. In the data-driven module (CNN-LSTM, Convolutional Neural Network - Long Short-Term Memory), the multi-dimensional feature vector constructed in step S202 is input into a hybrid model composed of a Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM). The CNN layer uses a one-dimensional convolution kernel to perform convolution operations on the multi-dimensional feature vector, extracting local spatial correlation features between features, and reduces the data dimensionality through pooling layers to reduce computational load. The LSTM layer receives the output of the CNN layer and uses its gating mechanism to capture the long-term dependencies and evolution patterns of battery fault data over time, outputting a preliminary time-series feature vector. In the knowledge-guided module, electrochemical physical mechanism features are calculated based on multi-level operational data, including charge / discharge coulombic efficiency, the degree of voltage deviation from open circuit voltage (OCV) curves, and battery capacity decay rate, constructing a physical feature vector. Subsequently, the temporal feature vector output by the CNN-LSTM module and the physical feature vector generated by the knowledge-guided module are concatenated and fused at the feature level to form a fused feature vector. Finally, the fused feature vector is input into the Gradient Boosting Decision Tree (GBDT) ensemble learning model, which outputs multi-granularity prediction results. These multi-granularity prediction results include binary classification results (determining whether a fault exists), continuous value prediction results (predicting fault severity indicators, such as the specific value of the maximum single-cell pressure difference), and trend prediction results (predicting the short-term trend of key indicators, such as rising, falling, or stabilizing). This data-driven and physical mechanism-integrated architecture leverages the powerful nonlinear fitting capabilities of deep learning models while incorporating electrochemical expert knowledge, improving the model's generalization ability and diagnostic accuracy under complex operating conditions, and providing a foundation for subsequent interpretability analysis.

[0041] Step S205: Based on the first fault detection result and the second fault detection result, determine the target fault detection result of the target battery system.

[0042] In this step, the main channel is a fault diagnosis channel based on dynamic thresholds, and the auxiliary channel is an intelligent diagnosis channel based on a hybrid-driven prediction model. Specifically, the first fault detection result output in step S203 is first obtained. If the first fault detection result is "normal," the target battery system is directly determined to be in normal condition, ending the current cycle's warning process. If the first fault detection result is "warning" or "alarm," the auxiliary channel verification mechanism is triggered. At this time, the binary classification result in the second fault detection result output in step S204 is read to determine whether it matches the warning or alarm status in the first fault detection result (i.e., whether both indicate a fault risk). If they match, it is confirmed as a valid warning, and the first and second fault detection results are integrated to generate the target fault detection result. If they do not match (e.g., the dynamic threshold channel triggers an alarm but the model determines it to be normal, or vice versa), a comprehensive judgment is made by combining continuous value prediction results (such as specific differential pressure values) and trend prediction results (such as whether the trend is upward or downward). The channel result with higher confidence or clearer risk is prioritized, or maintenance personnel are involved in the review to avoid false alarms or missed alarms. This strategy of combining main channel triggering with auxiliary channel verification leverages the high sensitivity of dynamic thresholds under specific operating conditions and the strong discriminative ability of ensemble learning models in complex feature spaces, achieving dual confirmation. This effectively reduces the false alarm rate, improves the accuracy of early warnings, and provides maintenance personnel with fine-grained fault situation awareness and interpretable decision-making basis by outputting continuous values ​​and trend information.

[0043] Through the above steps, the goal of combining multi-source heterogeneous data across levels, dynamic threshold adaptive adjustment and hybrid driving prediction model is achieved, and the collaborative decision-making of main channel triggering and auxiliary channel verification is realized. This achieves the technical effects of early and accurate warning, hierarchical diagnosis and high interpretability of results for energy storage battery faults. It also solves the technical problems of failing to capture early signals of fault propagation from individual cells to battery packs and battery clusters due to the single and isolated data source, lack of cross-scale correlation analysis and the traditional method's reliance on individual data.

[0044] As an optional implementation, based on multi-level operational data, a first feature, a second feature, and a third feature are extracted and concatenated into a multi-dimensional feature vector. This includes: calculating the voltage change rate and temperature change rate of each individual battery cell within a preset acquisition period based on their individual voltages and temperatures, thus constructing the first feature; calculating the mean voltage, standard deviation voltage, mean temperature, and standard deviation temperature of each individual battery cell within a preset acquisition period, based on their individual voltages and temperatures, thus constructing the second feature; calculating the maximum single-cell voltage difference of each battery pack based on the extreme values ​​of individual cell voltages; calculating the average voltage standard deviation of all battery packs within the same battery cluster; calculating the voltage-temperature correlation coefficient of each individual battery cell; and constructing the third feature based on the maximum single-cell voltage difference of each battery pack, the average voltage standard deviation of the battery pack, and the voltage-temperature correlation coefficient of each individual battery cell. The first, second, and third features are then concatenated and fused to obtain the multi-dimensional feature vector.

[0045] Optionally, firstly, the first derivatives (rates of change) of the voltage and temperature sequences of individual cells are extracted using a sliding window technique to construct a first feature reflecting the transient changes of the fault. Secondly, the mean, standard deviation, and maximum temperature difference within the sliding window are calculated to construct a second feature characterizing the steady-state distribution of the battery state. Next, the pack-level consistency feature is verified by calculating the maximum single-cell voltage difference across all cells within the same pack.

[0046]

[0047] in, The voltage of a single cell.

[0048] To verify the rack-level consistency characteristics, the average voltage standard deviation of all packs within the same rack was calculated:

[0049]

[0050] in, This represents the number of all packs within the same rack. Let be the average voltage of the i-th battery pack. This represents the average voltage of all battery packs within the battery rack.

[0051] Finally, to verify the thermoelectric coupling characteristics, the Pearson correlation coefficient between the voltage and temperature sequences was calculated:

[0052]

[0053] in, Voltage sequence data for a single cell. Temperature sequence data for a single cell.

[0054] The three calculation results together construct a third feature characterizing the correlation between individual battery cells, battery packs, and battery clusters. Finally, the first, second, and third features are concatenated and fused to form a multi-dimensional feature vector. This vector not only preserves transient and steady-state information at the individual cell level but also reveals early signals of fault propagation from individual cells to battery packs and battery clusters through cross-level correlation features. This provides a standardized input with both temporal evolution and spatial correlation characteristics for subsequent dynamic threshold determination and hybrid drive prediction models, thereby significantly improving the early detection capability and warning accuracy of energy storage battery faults.

[0055] As an optional embodiment, the first fault detection result of the target battery system is determined based on multi-level operating data, including: acquiring the historical standard voltage of the battery cells in the target battery system; determining a first coefficient based on the state of charge of each of the multiple battery cells; determining a second coefficient based on the temperature of each of the multiple battery cells; determining a third coefficient based on the total current of each of the multiple battery packs; determining a fourth coefficient based on the health status of each of the multiple battery cells; calculating the product of the historical standard voltage, the first coefficient, the second coefficient, the third coefficient, and the fourth coefficient to obtain a voltage difference threshold; and comparing the voltage difference threshold with the maximum single-cell voltage difference of each of the multiple battery packs to obtain the first fault detection result.

[0056] Optionally, firstly, based on the magnitude and direction of the real-time current I of the collected total pack current, it is determined whether the current is in a charging state (I<0), a discharging state (I>0), or a static state (I=0). Then, an adaptive threshold adjustment function is constructed, introducing a state-of-charge correction coefficient, a temperature correction coefficient, an operating condition correction coefficient, and a health correction coefficient to perform multi-dimensional coupled adjustment of the basic threshold. The dynamic threshold calculation formula is as follows:

[0057] Differential pressure threshold:

[0058] in, This is a statistical analysis of historical normal operation data. This is the state of charge correction factor. This is the temperature correction factor. This is the working condition correction factor. For health correction factors;

[0059] Temperature rate threshold:

[0060] in, This is a statistical analysis of historical normal operation data. The operating condition correction coefficients are as follows: the state of charge correction coefficient is greater than 1 when the state of charge is in the extreme range to relax the threshold; the temperature correction coefficient is greater than 1 when the temperature exceeds the safe range to reflect the aging acceleration effect; and the operating condition correction coefficient is greater than 1 during charging and discharging to consider dynamic stress. Finally, the real-time monitored battery characteristic values ​​are compared with the dynamically adjusted thresholds. Based on the degree of deviation and duration, the battery status is divided into three levels: normal, warning, or alarm. When the real-time value is less than the warning threshold, it is judged as a normal state. When the real-time value is greater than or equal to the warning threshold and exceeds the threshold continuously for a first preset time, a warning state is triggered. When the real-time value is greater than or equal to the fault threshold and exceeds the threshold continuously for a second preset time, an alarm state is triggered.

[0061] As an optional embodiment, determining the second fault detection result of the target battery system based on multidimensional feature vectors includes: making preliminary predictions on the fault probability and operating data of the target battery system based on multidimensional feature vectors to form a time-series feature vector; calculating the physical mechanism characteristics of individual battery cells in the target battery system based on multi-level operating data to form a physical feature vector; and fusing the time-series feature vector and the physical feature vector and inputting them into a preset gradient boosting decision tree ensemble learning model to obtain the second fault detection result.

[0062] Optionally, firstly, based on multi-dimensional feature vectors, the input layer of the data-driven model receives multi-dimensional feature vectors; the CNN layer uses a one-dimensional convolutional layer to perform convolution operations on the input multi-dimensional feature vectors, extracting local and deep spatial correlation features between features; then, the pooling layer reduces the data dimensionality and computational cost; finally, the LSTM layer takes the output of the CNN layer as input, utilizing the unique gating mechanism of LSTM to capture the long-term dependencies and evolution patterns of battery fault data over time; the output layer outputs preliminary fault probabilities and time-series predictions. Simultaneously, a knowledge-guided module is constructed: electrochemical knowledge is encoded into feature engineering rules to construct a physical mechanism feature set, including but not limited to:

[0063] Charge / discharge coulombic efficiency:

[0064]

[0065] in, This refers to the amount of charge discharged from a single battery cell during one charge-discharge cycle. This refers to the amount of charge a single battery cell receives during one charge-discharge cycle.

[0066] Degree of voltage deviation from the OCV curve:

[0067]

[0068] in, This represents the measured voltage of a single battery cell. This is the reference value for the open-circuit voltage at the current SOC.

[0069] Furthermore, the output features of the CNN-LSTM module are concatenated with the physical mechanism features generated by the knowledge-guided module to form a fused feature vector. This vector is then fed into the Gradient Boosting Decision Tree (GBDT) ensemble learning model for final decision-making, resulting in a multi-granularity prediction result that includes binary classification results, continuous value prediction results, and trend prediction results. The binary classification result indicates whether the target battery system has a fault, the continuous value prediction result indicates the severity of the fault, and the trend prediction result indicates the changing trend of the target battery system's operating data.

[0070] As an optional embodiment, determining the target fault detection result of the target battery system based on the first fault detection result and the second fault detection result includes: if the first fault detection result is a warning, determining whether the binary classification result of the first fault detection result and the second fault detection result matches; if the binary classification result of the first fault detection result and the second fault detection result matches, integrating the first fault detection result and the second fault detection result to obtain the target fault detection result.

[0071] Optionally, firstly, a first fault detection result based on a dynamic threshold mechanism is obtained. This result depends on the deviation between real-time operating data and an adaptive threshold, reflecting the immediate abnormal risk of the battery state under changing operating conditions. Simultaneously, a second fault detection result based on a hybrid-driven prediction model is obtained. This result depends on the fusion of extracted multi-dimensional spatiotemporal features and physical mechanism features, reflecting the probability of identifying deep battery fault modes. Consistency checks are performed on both results using logical judgment or a weighted voting mechanism. When both indicate a fault risk, the target fault detection result is confirmed as a high-confidence fault warning. If there is a conflict, the more interpretable physical mechanism feature analysis result is retained, or a comprehensive judgment is made by combining trend prediction results, to eliminate false alarms that may arise from single data-driven or single-threshold judgments.

[0072] In conjunction with the above optional embodiments, a method for early warning of energy storage battery faults based on multi-source data fusion and dynamic thresholds is also proposed. Figure 6 This is a general flowchart of a method for early warning of energy storage battery faults based on multi-source data fusion and dynamic thresholds according to an optional embodiment of the present invention, as shown below. Figure 6As shown, firstly, multi-level operational data of the energy storage battery is acquired; based on the multi-level operational data, time-series features, statistical features, and cross-level correlation features are extracted to form a multi-dimensional feature vector; based on the multi-dimensional feature vector, the fault judgment threshold is dynamically adjusted according to the battery's operating state, temperature, and state of charge to achieve graded early warning; based on the multi-dimensional feature vector, a hybrid-driven prediction framework is constructed, which integrates the spatiotemporal features extracted by CNN-LSTM with the electrochemical physical mechanism features and inputs them into an ensemble learning model to output multi-granularity prediction results; based on the graded early warning results and multi-granularity prediction results, a collaborative strategy of main channel triggering and auxiliary channel verification is adopted. When the dynamic threshold channel triggers an early warning, the binary classification results of the ensemble learning model are used for double confirmation, and the alarm information is refined by combining continuous values ​​and trend predictions.

[0073] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0074] Through the above description of the embodiments, those skilled in the art can clearly understand that the battery system fault detection method, apparatus, and non-volatile storage medium according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0075] According to an embodiment of the present invention, a battery system fault detection device for implementing the above-described battery system is also provided. Figure 7 This is a structural block diagram of a battery system fault detection device provided according to an embodiment of the present invention, such as... Figure 7 As shown, the battery system fault detection device includes: an acquisition module 71, an extraction module 72, a first determination module 73, a second determination module 74, and a third determination module 75. The battery system fault detection device will be described below.

[0076] The acquisition module 71 is used to acquire multi-level operating data of the target battery system at multiple preset sampling time points within a preset acquisition period. The multi-level operating data includes the operating data of multiple individual battery cells, the operating data of multiple battery packs, and the operating data of multiple battery clusters. Multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster.

[0077] The extraction module 72, connected to the acquisition module 71, is used to extract the first feature, the second feature, and the third feature based on multi-level operating data, and concatenate them into a multi-dimensional feature vector. The first feature represents the transient change characteristics of the operating data of the battery cell, the second feature represents the steady-state distribution characteristics of the operating data of the battery cell, and the third feature represents the correlation characteristics between the battery cell, the battery pack, and the battery cluster.

[0078] The first determining module 73, connected to the extraction module 72, is used to determine the first fault detection result of the target battery system based on multi-level operating data. The first fault detection result includes whether the target battery system is in a normal, warning, or alarm state.

[0079] The second determining module 74, connected to the first determining module 73, is used to determine the second fault detection result of the target battery system based on multi-dimensional feature vectors and multi-level operating data. The second fault detection result includes binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification results indicate whether the target battery system has a fault, the continuous value prediction results indicate the severity index of the fault, the trend prediction results indicate the changing trend of the operating data of the target battery system, and the list of key influencing factors indicates the cause of the fault in the target battery system.

[0080] The third determining module 75, connected to the second determining module 74, is used to determine the target fault detection result of the target battery system based on the first fault detection result and the second fault detection result.

[0081] It should be noted that the aforementioned acquisition module 71, extraction module 72, first determination module 73, second determination module 74, and third determination module 75 correspond to steps S201 to S205 in the embodiments. Multiple modules implement the same instances and application scenarios as their corresponding steps, but are not limited to the content disclosed in the above embodiments. It should also be noted that the aforementioned modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0082] Embodiments of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0083] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the energy storage battery fault early warning method and device based on multi-source data fusion and dynamic thresholds in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned energy storage battery fault early warning method based on multi-source data fusion and dynamic thresholds. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0084] The processor can invoke information and application programs stored in the memory via a transmission device to perform the following steps: acquiring multi-level operational data of the target battery system at multiple preset sampling time points, wherein the multi-level operational data includes operational data of multiple individual battery cells, operational data of multiple battery packs, and operational data of multiple battery clusters; extracting a first feature, a second feature, and a third feature based on the multi-level operational data, and concatenating them into a multi-dimensional feature vector, wherein the first feature characterizes the transient change characteristics of the operational data of individual battery cells, the second feature characterizes the steady-state distribution characteristics of the operational data of individual battery cells, and the third feature characterizes the correlation characteristics between individual battery cells, battery packs, and battery clusters; determining a first fault detection result of the target battery system based on the multi-level operational data, wherein the first fault detection result includes the target battery system's status as normal, warning, or alarm; determining a second fault detection result of the target battery system based on the multi-dimensional feature vector and the multi-level operational data, wherein the second fault detection result includes binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors; and determining a target fault detection result of the target battery system based on the first fault detection result and the second fault detection result.

[0085] Optionally, the processor may also execute program code that performs the following steps: within a preset acquisition period, acquire multi-level operating data of the target battery system at multiple preset sampling time points, including: the voltage of each of the multiple battery cells, the temperature of each of the multiple battery cells, the state of charge of each of the multiple battery cells, the health status of each of the multiple battery cells, the total current of each of the multiple battery packs, the extreme values ​​of the individual cell voltages of each of the multiple battery packs, the maximum temperature difference of each of the multiple battery packs, the total voltage of each of the multiple battery clusters, the total current of each of the multiple battery clusters, and the ambient temperature of each of the multiple battery clusters.

[0086] Optionally, the processor may also execute program code with the following steps: Based on multi-level operational data, extract a first feature, a second feature, and a third feature to form a multi-dimensional feature vector, including: based on the voltage and temperature of each of the multiple battery cells, calculate the voltage change rate and temperature change rate of each of the multiple battery cells within a preset acquisition period to construct the first feature; based on the voltage and temperature of each of the multiple battery cells, calculate the mean voltage, standard deviation voltage, mean temperature, and standard deviation temperature of each of the multiple battery cells within a preset acquisition period to construct the second feature; based on the extreme values ​​of the individual cell voltages of the multiple battery packs, calculate the maximum individual cell voltage difference of each of the multiple battery packs; calculate the average voltage standard deviation of all battery packs in the same battery cluster; calculate the voltage-temperature correlation coefficient of each of the multiple battery cells; and construct the third feature based on the maximum individual cell voltage difference of each of the multiple battery packs, the average voltage standard deviation of the battery pack, and the voltage-temperature correlation coefficient of each of the multiple battery cells.

[0087] Optionally, the processor may also execute program code for the following steps: determining a first fault detection result of the target battery system based on multi-level operational data, including: acquiring the historical standard voltage of a single battery cell in the target battery system; determining a first coefficient based on the state of charge of multiple single battery cells; determining a second coefficient based on the temperature of multiple single battery cells; determining a third coefficient based on the total current of multiple battery packs; determining a fourth coefficient based on the health status of multiple single battery cells; calculating the product of the historical standard voltage, the first coefficient, the second coefficient, the third coefficient, and the fourth coefficient to obtain a voltage difference threshold; and comparing the voltage difference threshold with the maximum single-cell voltage difference of multiple battery packs to obtain the first fault detection result.

[0088] Optionally, the processor may also execute program code for the following steps: determining the second fault detection result of the target battery system based on multi-dimensional feature vectors, including: making preliminary predictions on the fault probability and operating data of the target battery system based on multi-dimensional feature vectors to form a time-series feature vector; calculating the physical mechanism characteristics of individual battery cells in the target battery system based on multi-level operating data to form a physical feature vector; and fusing the time-series feature vector and the physical feature vector and inputting them into a preset gradient boosting decision tree ensemble learning model to obtain the second fault detection result.

[0089] Optionally, the processor may also execute program code for the following steps: determining the target fault detection result of the target battery system based on the first fault detection result and the second fault detection result, including: if the first fault detection result is a warning, determining whether the binary classification result of the first fault detection result and the second fault detection result matches; if the binary classification result of the first fault detection result and the second fault detection result matches, integrating the first fault detection result and the second fault detection result to obtain the target fault detection result.

[0090] This invention provides a battery system fault detection method. It acquires multi-level operational data of a target battery system at multiple preset sampling time points within a preset acquisition period. This multi-level operational data includes operational data for individual battery cells, battery packs, and battery clusters. Multiple battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster. Based on the multi-level operational data, a first feature, a second feature, and a third feature are extracted to form a multi-dimensional feature vector. The first feature represents the transient change characteristics of the operational data of individual battery cells, the second feature represents the steady-state distribution characteristics of the operational data of individual battery cells, and the third feature represents the correlation characteristics between battery cells, battery packs, and battery clusters. Based on the multi-level operational data, a first fault detection result of the target battery system is determined, including whether the target battery system is in normal, warning, or alarm state. Based on the multi-dimensional feature vector and the multi-level operational data, a second fault detection result of the target battery system is determined. The fault detection results include a second fault detection result comprising binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification results indicate whether a fault exists in the target battery system, the continuous value prediction results indicate the severity of the fault, the trend prediction results indicate the changing trend of the target battery system's operating data, and the list of key influencing factors indicates the cause of the fault in the target battery system. Based on the first and second fault detection results, the target fault detection result of the target battery system is determined. This achieves the goal of triggering early warning through a dynamic threshold channel and using a hybrid driving model for dual confirmation and fine-grained information supplementation, realizing the collaborative decision-making of main channel triggering and auxiliary channel verification. This achieves the technical effects of early and accurate early warning, graded diagnosis, and high interpretability of results for energy storage battery faults. Furthermore, it solves the technical problem that traditional methods, which rely solely on individual data and lack cross-scale correlation analysis due to the single and isolated data source, cannot capture early signals of fault propagation from individual cells to battery packs and battery clusters.

[0091] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0092] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the energy storage battery fault early warning method based on multi-source data fusion and dynamic threshold provided in the above embodiments.

[0093] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0094] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: within a preset acquisition period, acquiring multi-level operational data of the target battery system at multiple preset sampling time points, wherein the multi-level operational data includes operational data of multiple individual battery cells, operational data of multiple battery packs, and operational data of multiple battery clusters, wherein multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster; based on the multi-level operational data, extracting a first feature, a second feature, and a third feature to form a multi-dimensional feature vector, wherein the first feature characterizes the transient change characteristics of the operational data of individual battery cells, the second feature characterizes the steady-state distribution characteristics of the operational data of individual battery cells, and the third feature characterizes the correlation characteristics between individual battery cells, battery packs, and battery clusters. The system is further divided into two parts: First, based on multi-level operational data, a first fault detection result is determined for the target battery system, including whether the target battery system is in normal, warning, or alarm state. Second, based on multi-dimensional feature vectors and multi-level operational data, a second fault detection result is determined for the target battery system, including binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification results indicate whether a fault exists in the target battery system, the continuous value prediction results indicate the severity of the fault, the trend prediction results indicate the changing trend of the target battery system's operational data, and the list of key influencing factors indicates the cause of the fault in the target battery system. Finally, based on the first and second fault detection results, a target fault detection result for the target battery system is determined.

[0095] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: within a preset acquisition period, acquiring multi-level operating data of the target battery system at multiple preset sampling time points, including: the voltage of each of the multiple battery cells, the temperature of each of the multiple battery cells, the state of charge of each of the multiple battery cells, the health status of each of the multiple battery cells, the total current of each of the multiple battery packs, the extreme values ​​of the individual cell voltages of each of the multiple battery packs, the maximum temperature difference of each of the multiple battery packs, the total voltage of each of the multiple battery clusters, the total current of each of the multiple battery clusters, and the ambient temperature of each of the multiple battery clusters.

[0096] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: extracting a first feature, a second feature, and a third feature based on multi-level operational data to form a multi-dimensional feature vector, including: calculating the voltage change rate and temperature change rate of each of the multiple battery cells within a preset acquisition period based on the voltage and temperature of each of the multiple battery cells, and constructing the first feature; calculating the average voltage, voltage standard deviation, average temperature, and temperature standard deviation of each of the multiple battery cells within a preset acquisition period based on the voltage and temperature of each of the multiple battery cells, and constructing the second feature; calculating the maximum single-cell voltage difference of each of the multiple battery packs based on the extreme values ​​of the single-cell voltages of each of the multiple battery packs; calculating the average voltage standard deviation of all battery packs in the same battery cluster; calculating the voltage-temperature correlation coefficient of each of the multiple battery cells; and constructing the third feature based on the maximum single-cell voltage difference of each of the multiple battery packs, the average voltage standard deviation of the battery pack, and the voltage-temperature correlation coefficient of each of the multiple battery cells.

[0097] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining a first fault detection result of the target battery system based on multi-level operating data, including: acquiring the historical standard voltage of a single battery cell in the target battery system; determining a first coefficient based on the state of charge of multiple battery cells; determining a second coefficient based on the temperature of multiple battery cells; determining a third coefficient based on the total current of multiple battery packs; determining a fourth coefficient based on the health status of multiple battery cells; calculating the product of the historical standard voltage, the first coefficient, the second coefficient, the third coefficient, and the fourth coefficient to obtain a voltage difference threshold; and comparing the voltage difference threshold with the maximum single-cell voltage difference of multiple battery packs to obtain the first fault detection result.

[0098] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining a second fault detection result of the target battery system based on multi-dimensional feature vectors, including: making preliminary predictions on the fault probability and operating data of the target battery system based on multi-dimensional feature vectors to form a time-series feature vector; calculating the physical mechanism characteristics of individual battery cells in the target battery system based on multi-level operating data to form a physical feature vector; and fusing the time-series feature vector and the physical feature vector and inputting them into a preset gradient boosting decision tree ensemble learning model to obtain the second fault detection result.

[0099] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the target fault detection result of the target battery system based on the first fault detection result and the second fault detection result, including: if the first fault detection result is a warning, determining whether the binary classification result of the first fault detection result and the second fault detection result matches; if the binary classification result of the first fault detection result and the second fault detection result matches, integrating the first fault detection result and the second fault detection result to obtain the target fault detection result.

[0100] Embodiments of the present invention also provide a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can: acquire multi-level operational data of a target battery system at multiple preset sampling time points within a preset acquisition period, wherein the multi-level operational data includes operational data of multiple individual battery cells, operational data of multiple battery packs, and operational data of multiple battery clusters, wherein multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster; based on the multi-level operational data, extract a first feature, a second feature, and a third feature, and concatenate them into a multi-dimensional feature vector, wherein the first feature characterizes the transient change characteristics of the operational data of individual battery cells, the second feature characterizes the steady-state distribution characteristics of the operational data of individual battery cells, and the third feature characterizes the battery cells and battery packs. The system identifies the correlation characteristics between the battery cluster and the target battery system. Based on multi-level operational data, it determines the first fault detection result of the target battery system, which includes whether the target battery system is in normal, warning, or alarm state. Based on multi-dimensional feature vectors and multi-level operational data, it determines the second fault detection result of the target battery system, which includes binary classification results, continuous value prediction results, trend prediction results, and a list of key influencing factors. The binary classification results indicate whether the target battery system has a fault, the continuous value prediction results indicate the severity of the fault, the trend prediction results indicate the changing trend of the target battery system's operational data, and the list of key influencing factors indicates the cause of the fault in the target battery system. Based on the first and second fault detection results, it determines the target fault detection result of the target battery system.

[0101] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0102] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0104] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0105] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0106] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, 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 network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0107] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting battery system faults, characterized in that, include: Within a preset acquisition period, multi-level operational data of the target battery system at multiple preset sampling time points are acquired. The multi-level operational data includes the operational data of multiple individual battery cells, the operational data of multiple battery packs, and the operational data of multiple battery clusters. Multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster. Based on the multi-level operating data, a first feature, a second feature, and a third feature are extracted and concatenated into a multi-dimensional feature vector. The first feature represents the transient change characteristics of the operating data of the battery cell, the second feature represents the steady-state distribution characteristics of the operating data of the battery cell, and the third feature represents the correlation characteristics between the battery cell, the battery pack, and the battery cluster. Based on the multi-level operating data, a first fault detection result of the target battery system is determined, wherein the first fault detection result includes whether the state of the target battery system is normal, warning, or alarm. Based on the multidimensional feature vector and the multi-level operating data, a second fault detection result of the target battery system is determined. The second fault detection result includes a binary classification result, a continuous value prediction result, a trend prediction result, and a list of key influencing factors. The binary classification result indicates whether the target battery system has a fault. The continuous value prediction result indicates the severity index of the fault. The trend prediction result indicates the changing trend of the operating data of the target battery system. The list of key influencing factors indicates the cause of the fault in the target battery system. Based on the first fault detection result and the second fault detection result, the target fault detection result of the target battery system is determined.

2. The method according to claim 1, characterized in that, The operating data of each individual battery cell includes the voltage of each individual battery cell, the temperature of each individual battery cell, the state of charge of each individual battery cell, the health status of each individual battery cell, the total current of each battery pack, the extreme voltage of each individual cell in each battery pack, the maximum temperature difference of each battery pack, the total voltage of each battery cluster, the total current of each battery cluster, and the ambient temperature of each battery cluster.

3. The method according to claim 2, characterized in that, The step of extracting the first feature, the second feature, and the third feature based on the multi-level operational data, and concatenating them into a multi-dimensional feature vector, includes: Based on the voltage and temperature of each of the multiple battery cells, the voltage change rate and temperature change rate of each of the multiple battery cells within the preset acquisition period are calculated and used as the first feature. Based on the voltage and temperature of each of the multiple battery cells, the average voltage, standard deviation of voltage, average temperature, and standard deviation of temperature of each of the multiple battery cells within the preset acquisition period are calculated and used as the second feature. Based on the extreme values ​​of the individual cell voltages of the plurality of battery packs, calculate the maximum individual cell voltage difference of the plurality of battery packs; Calculate the average voltage standard deviation of all battery packs in the same battery cluster; Calculate the voltage-temperature correlation coefficient for each of the multiple battery cells; The third feature is defined as the maximum single-cell voltage difference of each of the plurality of battery packs, the average voltage standard deviation of the battery packs, and the voltage-temperature correlation coefficient of each of the plurality of battery cells.

4. The method according to claim 2, characterized in that, The determination of the first fault detection result of the target battery system based on the multi-level operational data includes: Obtain the historical standard voltage of the battery cells in the target battery system; A first coefficient is determined based on the state of charge of each of the multiple battery cells; The second coefficient is determined based on the temperature of each of the multiple battery cells; The third coefficient is determined based on the total current of each of the multiple battery packs; A fourth coefficient is determined based on the health status of each of the multiple battery cells; The voltage difference threshold is obtained by calculating the product of the historical standard voltage, the first coefficient, the second coefficient, the third coefficient, and the fourth coefficient. The first fault detection result is obtained by comparing the voltage difference threshold with the maximum single-cell voltage difference of each of the multiple battery packs.

5. The method according to claim 1, characterized in that, The determination of the second fault detection result of the target battery system based on the multidimensional feature vector includes: Based on the multidimensional feature vector, a preliminary prediction is made of the failure probability and operating data of the target battery system, forming a time-series feature vector. Based on the multi-level operational data, the physical mechanism characteristics of the battery cells in the target battery system are calculated to form a physical feature vector; The time-series feature vector and the physical feature vector are fused and then input into a preset gradient boosting decision tree ensemble learning model to obtain the second fault detection result.

6. The method according to claim 1, characterized in that, Determining the target fault detection result of the target battery system based on the first fault detection result and the second fault detection result includes: If the first fault detection result is a warning, determine whether the binary classification results of the first fault detection result and the second fault detection result match; If the binary classification results of the first fault detection result and the second fault detection result match, the first fault detection result and the second fault detection result are integrated to obtain the target fault detection result.

7. A battery system fault detection device, characterized in that, include: The acquisition module is used to acquire multi-level operating data of the target battery system at multiple preset sampling time points within a preset acquisition period. The multi-level operating data includes the operating data of multiple individual battery cells, the operating data of multiple battery packs, and the operating data of multiple battery clusters. Multiple individual battery cells constitute a battery pack, and multiple battery packs constitute a battery cluster. The extraction module is used to extract a first feature, a second feature, and a third feature based on the multi-level operating data, and concatenate them into a multi-dimensional feature vector. The first feature represents the transient change characteristics of the operating data of the battery cell, the second feature represents the steady-state distribution characteristics of the operating data of the battery cell, and the third feature represents the correlation characteristics between the battery cell, the battery pack, and the battery cluster. The first determining module is used to determine the first fault detection result of the target battery system based on the multi-level operating data, wherein the first fault detection result includes the state of the target battery system as normal, warning, or alarm. The second determining module is used to determine a second fault detection result of the target battery system based on the multidimensional feature vector and the multi-level operating data. The second fault detection result includes a binary classification result, a continuous value prediction result, a trend prediction result, and a list of key influencing factors. The binary classification result indicates whether the target battery system has a fault. The continuous value prediction result indicates the severity index of the fault. The trend prediction result indicates the changing trend of the operating data of the target battery system. The list of key influencing factors indicates the cause of the fault in the target battery system. The third determining module is used to determine the target fault detection result of the target battery system based on the first fault detection result and the second fault detection result.

8. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the non-volatile storage medium to perform the battery system fault detection method according to any one of claims 1 to 6.

9. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor performs the battery system fault detection method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the battery system fault detection method according to any one of claims 1 to 6.

Images