Energy storage system based on multi-source information fusion and composite anomaly analysis, comprehensive fault prediction method, equipment and medium
By integrating multi-source information and performing composite anomaly analysis, a baseline model and a multi-source information fusion model are established to calculate the comprehensive anomaly probability value (CAI) of the energy storage system. This solves the problem of insufficient perception and trend analysis of composite anomalies in existing technologies, and enables early fault warning and risk assessment.
Patent Information
- Application Number
- CN202511742131.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-25
- Publication Date
- 2026-03-06
AI Technical Summary
Existing energy storage systems cannot effectively detect, quantify, and analyze trends under complex anomalies, resulting in high false alarm rates and a lack of early warning capabilities, making them unable to cope with the risk of thermal runaway in lithium-ion batteries.
By employing a multi-source information fusion and composite anomaly analysis method, a baseline model and a multi-source information fusion model are established by acquiring basic parameters and historical operational data from multiple sources. The comprehensive anomaly probability value (CAI) is calculated, and different levels of early warnings are issued based on the CAI curve.
It enables early fault prediction and risk level classification of energy storage systems, reduces false alarm rate, improves the accuracy and adaptability of early warning, and can provide early warning of serious faults such as thermal runaway.
Smart Images
Figure CN121615072A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of energy storage data processing technology, and more specifically, to an energy storage system and a comprehensive fault prediction method, device, and medium based on multi-source information fusion and composite anomaly analysis. Background Technology
[0002] With the acceleration of the global energy transition, the electrochemical energy storage market continues to expand. However, lithium-ion batteries are susceptible to thermal runaway under conditions such as overcharging, over-discharging, internal short circuits, mechanical abuse, or thermal abuse, which may lead to fires or even explosions, seriously threatening life and property safety. In recent years, frequent safety accidents at energy storage power stations both domestically and internationally have made "safety" the most critical factor restricting the industry's development.
[0003] Current predictive systems suffer from rigidity and high false alarm rates, relying on static thresholds or simple Boolean logic for their core logic. Energy storage systems operate under complex and variable conditions; during high-rate charging and discharging, temperature and voltage fluctuations increase significantly. Static thresholds cannot adapt to these dynamic changes, easily generating numerous false alarms during periods of fluctuating operating conditions, while failing to trigger alarms in truly dangerous composite anomalies where individual parameters do not exceed thresholds.
[0004] Lack of quantification and trend judgment: This solution can only make binary "yes / no" judgments, and cannot quantify the degree of anomalies, let alone perceive the trend of multiple weak anomalies deteriorating together over time. It can only be triggered when the anomaly has become very obvious, and is essentially still a lagging alarm rather than a predictive warning.
[0005] In summary, existing technologies are still unable to effectively perceive, quantify, and analyze trends in "compound anomalies"—that is, the simultaneous occurrence of weak, secondary abnormal changes in multiple parameters. They are either too narrow in perspective or too rigid in logic, making it difficult to provide reliable and accurate early warnings in the very early stages of an accident. Summary of the Invention The purpose of this invention is to provide an energy storage system and a comprehensive fault prediction method, device, and medium based on multi-source information fusion and composite anomaly analysis to solve the above-mentioned problems in the prior art.
[0006] This invention is achieved through the following technical solution: In a first aspect, the present invention provides a comprehensive fault prediction method for energy storage systems based on multi-source information fusion and composite anomaly analysis, comprising: Acquire basic parameters and historical operational data from multiple sources, and preprocess the historical operational data and basic parameters for each type of parameter; A baseline model is established based on historical operational data, and anomaly scores for each parameter are obtained through the baseline model based on the current basic parameters. A multi-source information fusion model is established, and the comprehensive anomaly probability value (CAI) at the current moment is output through the multi-source information fusion model based on the anomaly score of each parameter. The CAI curve is obtained by sorting all the currently collected comprehensive anomaly probability values according to the time sequence. Different levels of early warning are issued based on the comprehensive anomaly probability values and growth trends of the CAI curve.
[0007] Preferably, the preprocessing of historical operating data and basic parameters for each parameter includes: Missing and outlier values are removed from the historical operating data and basic parameters. The missing and outlier values are then supplemented by averaging the adjacent data. Normalize the historical running data and basic parameters for each parameter, and output the normalized historical running data and basic parameters;
[0008] In the formula, For the j-th parameter, the data at time i is normalized. For the data at time i of the j-th parameter, This is the minimum value of the current type parameter. This is the maximum value of the current category parameter; The normalized historical operating data and basic parameters are timestamped and then sorted according to the timestamps.
[0009] Preferably, the establishment of a baseline model based on historical operational data includes: Obtain historical operational data for each parameter, and establish a confidence interval for the maximum and minimum values at each time point. The confidence interval is the range of data encompassed in the time series. Connecting several maximum values forms the upper limit of the confidence interval on the time series, and connecting several minimum values forms the lower limit of the confidence interval on the time series. Data within the confidence interval is considered to be within the normal fluctuation range, and the upper and lower limits of the confidence interval represent the healthy baseline. The curve formed by connecting the average of the maximum and minimum values at several points in time constitutes the expected value curve.
[0010] Preferably, obtaining the outlier score for each parameter based on the current baseline parameters through the baseline model includes: Anomaly scores are obtained by calculating the deviation between the timestamp of the basic parameters and the expected value curve of the same type of parameters.
[0011] In the formula, For the anomaly score of the j-th basic parameter at time i, This represents the total number of time series. For the data at time i of the j-th basic parameter, This represents the data at time i of the expected value curve.
[0012] Preferably, the establishment of the multi-source information fusion model includes:
[0013] In the formula, Let be the comprehensive anomaly probability value of the current energy storage system at time i. The weights for calculating the j-th basic parameter are: The types of basic parameters; By connecting the combined anomaly probability values at all times, the CAI curve is obtained.
[0014] Preferably, the issuance of different levels of early warning based on the comprehensive anomaly probability value and growth trend of the CAI curve includes: The warning levels include Level 1, Level 2, and Level 3 warnings, each with its own threshold. When the overall anomaly probability value at time i is greater than or equal to the first-level threshold and less than the second-level threshold, a first-level warning is issued; when the overall anomaly probability value at time i is less than the first-level threshold, no warning is issued. When the comprehensive anomaly probability value at time i is greater than or equal to the level 2 threshold and less than the level 3 threshold, a level 2 warning is issued. When the comprehensive anomaly probability value at time i is greater than or equal to the level three threshold, a level three warning is issued.
[0015] Preferably, this includes establishing a transition model, which determines whether the current warning level should be changed when the issued warning is not a Level 3 warning; The transition model includes:
[0016] In the formula, For the over-limit rate, The number of time points when the basic data exceeds the upper or lower limit; when hour, The threshold for the transition model remains unchanged. At the current time point, obtain the r comprehensive anomaly probability values on the CAI curve after the current time point. Based on the r comprehensive anomaly probability values, determine whether the growth rate of the current comprehensive anomaly probability value is increasing. If it is increasing, modify the current warning level to the warning level of the previous level and output it. If it is not increasing, do not make any changes.
[0017] Secondly, the present invention also provides a comprehensive fault prediction system for energy storage systems based on multi-source information fusion and composite anomaly analysis, used to execute the aforementioned comprehensive fault prediction method for energy storage systems based on multi-source information fusion and composite anomaly analysis, comprising: The data processing module is configured to acquire basic parameters and historical operating data from multiple sources, preprocess the historical operating data and basic parameters for each parameter, establish a baseline model based on the historical operating data, and obtain the anomaly score for each parameter based on the current basic parameters through the baseline model. The prediction module is configured to establish a multi-source information fusion model, output the comprehensive anomaly probability value (CAI) at the current moment based on the anomaly score of each parameter through the multi-source information fusion model; sort all the currently collected comprehensive anomaly probability values according to the time sequence to obtain the CAI curve, and issue different levels of early warning based on the comprehensive anomaly probability value and growth trend of the CAI curve.
[0018] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-described method for comprehensive fault prediction of energy storage systems based on multi-source information fusion and composite anomaly analysis.
[0019] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for comprehensive fault prediction of energy storage systems based on multi-source information fusion and composite anomaly analysis.
[0020] The technical solution of the present invention has at least the following advantages and beneficial effects: The method provided by this invention mainly includes establishing a baseline model based on historical operating data, obtaining anomaly scores for each parameter based on current basic parameters through the baseline model, outputting a comprehensive anomaly probability value (CAI) for the current moment through a multi-source information fusion model based on the anomaly scores of each parameter, and issuing early warnings of different levels based on the comprehensive anomaly probability value and growth trend of the CAI curve. By collecting multi-dimensional operating data (electrical, thermal, chemical, and state) of the energy storage system, a dynamic health baseline is established for each parameter that changes with operating conditions, thereby detecting minor primary anomalies. Then, through an information fusion model, these primary anomalies are comprehensively calculated into a composite anomaly index that can globally reflect the health status of the system. Finally, by continuously tracking the time-series behavior of CAI and setting up a multi-level early warning mechanism, early prediction of faults and risk level classification are achieved. This addresses the lack of quantification and trend judgment in existing technologies. Attached Figure Description
[0021] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is a schematic diagram of the system structure of the present invention. Detailed Implementation
[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0024] The module divisions described in this application are logical divisions. In practical applications, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the connections, couplings, or communications in this application can be direct connections, couplings, or communications between related objects, or indirect connections, couplings, or communications through other devices. Moreover, the connections, couplings, or communications between objects can be electrical or other similar forms, and are not limited in this application.
[0025] The independently described modules or sub-modules may or may not be physically separated; they may be implemented in software or hardware, and some modules or sub-modules may be implemented in software, with the processor calling the software to implement the function of these modules or sub-modules, while other modules or sub-modules may be implemented in hardware, such as through hardware circuits. Furthermore, some or all of the modules can be selected to achieve the purpose of this application's solution according to actual needs.
[0026] Please refer to Figures 1-2 This invention provides a comprehensive fault prediction method for energy storage systems based on multi-source information fusion and composite anomaly analysis, comprising: S101: Obtain the basic parameters and historical operating data of multi-source information, and preprocess the historical operating data and basic parameters of each parameter; Data sources: Battery Management System (BMS), Energy Storage Converter (PCS), Environmental Monitoring Sensors (Temperature and Humidity), Fire Alarm System (Smoke, Combustible Gas <e.g., H2, CO, VOCs> Sensors), Thermal Management Control System (Coolant Flow Rate, Temperature).
[0027] Data collected: Electrical parameters: cell / cluster voltage, current, DC internal resistance, insulation resistance. Thermal parameters: cell surface temperature, inter-module temperature difference, battery compartment ambient temperature, coolant inlet and outlet temperatures. Chemical parameters: concentration of specific gases (H2, CO, VOCs), smoke concentration. State parameters: SOC, SOH, cumulative cycle count, operating time. Operating condition parameters: charge / discharge rate, ambient temperature.
[0028] S102: Establish a baseline model based on historical operating data, and obtain the anomaly score for each parameter based on the current basic parameters through the baseline model; For each key feature parameter, a model is trained based on a large amount of historical normal operation data to reflect its normal changes with operating parameters (such as SOC, current, and ambient temperature).
[0029] Model selection: For linear relationships, multiple linear regression can be used. For nonlinear relationships, machine learning algorithms such as Gaussian process regression (GPR) or XGBoost can be used.
[0030] Output: For any given operating condition at any given time, the model can predict the expected value and normal fluctuation range (confidence interval) of the characteristic parameter. This dynamic range is the healthy baseline.
[0031] Real-time data is input into each baseline model, and the deviation between the measured value and the expected value of each basic parameter is calculated. The higher the score, the further the parameter deviates from the normal range.
[0032] S103: Establish a multi-source information fusion model, and output the comprehensive anomaly probability value CAI at the current moment based on the anomaly score of each parameter through the multi-source information fusion model; S104: Sort all the currently collected comprehensive anomaly probability values according to the time sequence to obtain the CAI curve, and issue different levels of early warning based on the comprehensive anomaly probability value and growth trend of the CAI curve.
[0033] The method provided by this invention mainly includes establishing a baseline model based on historical operating data, obtaining anomaly scores for each parameter based on current basic parameters through the baseline model, outputting a comprehensive anomaly probability value (CAI) for the current moment through a multi-source information fusion model based on the anomaly scores of each parameter, and issuing early warnings of different levels based on the comprehensive anomaly probability value and growth trend of the CAI curve. By collecting multi-dimensional operating data (electrical, thermal, chemical, and state) of the energy storage system, a dynamic health baseline is established for each parameter that changes with operating conditions, thereby detecting minor primary anomalies. Then, through an information fusion model, these primary anomalies are comprehensively calculated into a composite anomaly index that can globally reflect the health status of the system. Finally, by continuously tracking the time-series behavior of CAI and setting up a multi-level early warning mechanism, early prediction of faults and risk level classification are achieved. This addresses the lack of quantification and trend judgment in existing technologies.
[0034] In one exemplary embodiment of the present invention, preprocessing the historical operating data and basic parameters for each parameter includes: Missing and outlier values are removed from the historical operating data and basic parameters. The missing and outlier values are then supplemented by averaging the adjacent data. Normalize the historical running data and basic parameters for each parameter, and output the normalized historical running data and basic parameters;
[0035] In the formula, For the j-th parameter, the data at time i is normalized. For the data at time i of the j-th parameter, This is the minimum value of the current type parameter. This is the maximum value of the current category parameter; The normalized historical operating data and basic parameters are timestamped and then sorted according to the timestamps.
[0036] An exemplary embodiment of the present invention, establishing a baseline model based on historical operational data, includes: Obtain historical operational data for each parameter, and establish a confidence interval for the maximum and minimum values at each time point. The confidence interval is the range of data encompassed in the time series. Connecting several maximum values forms the upper limit of the confidence interval on the time series, and connecting several minimum values forms the lower limit of the confidence interval on the time series. Data within the confidence interval is considered to be within the normal fluctuation range, and the upper and lower limits of the confidence interval represent the healthy baseline. The curve formed by connecting the average of the maximum and minimum values at several points in time constitutes the expected value curve.
[0037] In one exemplary embodiment of the present invention, obtaining the outlier score for each parameter through a baseline model based on the current base parameters includes: Anomaly scores are obtained by calculating the deviation between the timestamp of the basic parameters and the expected value curve of the same type of parameters.
[0038] In the formula, For the anomaly score of the j-th basic parameter at time i, This represents the total number of time series. For the data at time i of the j-th basic parameter, This represents the data at time i of the expected value curve.
[0039] An exemplary embodiment of the present invention includes establishing a multi-source information fusion model, comprising:
[0040] In the formula, Let be the comprehensive anomaly probability value of the current energy storage system at time i. The weights for calculating the j-th basic parameter are: The types of basic parameters; By connecting the combined anomaly probability values at all times, the CAI curve is obtained.
[0041] The weights can be determined by expert experience or model learning based on the importance of the parameters (e.g., gas concentration and temperature rise rate have higher weights).
[0042] Machine learning fusion model (preferred): The initial anomaly score vector is input into a classification model (such as Support Vector Machine, Random Forest, or Shallow Neural Network), which outputs a comprehensive anomaly probability value between 0 and 1 as the CAI. This model can be trained under supervision using historical fault data, resulting in better performance.
[0043] In one exemplary embodiment of the present invention, issuing warnings of different levels based on the comprehensive anomaly probability value and growth trend of the CAI curve includes: The warning levels include Level 1, Level 2, and Level 3 warnings, each with its own threshold. When the overall anomaly probability value at time i is greater than or equal to the first-level threshold and less than the second-level threshold, a first-level warning is issued; when the overall anomaly probability value at time i is less than the first-level threshold, no warning is issued. When the comprehensive anomaly probability value at time i is greater than or equal to the level 2 threshold and less than the level 3 threshold, a level 2 warning is issued. When the comprehensive anomaly probability value at time i is greater than or equal to the level three threshold, a level three warning is issued.
[0044] Specifically, Level 1 (Attention / Early Warning): CAI continuously exceeds the Level 1 threshold for a period of time. The system logs this information and displays a yellow alert on the monitoring interface, prompting operations and maintenance personnel to pay attention.
[0045] Level 2 (Advanced / Maintenance Warning): The CAI (Continuous Improvement Ability) rapidly exceeds a higher Level 2 threshold, or the CAI trend slope remains positive and the value is large. The system automatically sends SMS and email alerts to the operations and maintenance manager, recommending that an inspection be arranged.
[0046] Level 3 (Emergency / Immediate Hazard): The CAI (Critical Illness) exceeds the Level 3 threshold and is accompanied by a specific combination of characteristics (such as simultaneous drastic changes in temperature rise rate and CO concentration). The system immediately activates audible and visual alarms and can automatically execute linkage controls (such as forced shutdown and activation of the fire suppression system).
[0047] In one embodiment, a transition model is established, and when the issued warning is not a level 3 warning, the transition model is used to determine whether the current warning level has changed. The transition model includes:
[0048] In the formula, For the over-limit rate, The number of time points when the basic data exceeds the upper or lower limit; when hour, The threshold for judging the transition model can be set to 0.2 in this embodiment without modification. At the current time point, obtain the r comprehensive anomaly probability values on the CAI curve after the current time point. Based on the r comprehensive anomaly probability values, determine whether the growth rate of the current comprehensive anomaly probability value is increasing. If it is increasing, modify the current warning level to the warning level of the previous level and output it. If it is not increasing, do not make any changes.
[0049] By using a transition model, more accurate early warnings can be issued. For example, if the current warning is at level one, it can be changed to level two when a transition is needed.
[0050] Once the warning is triggered, the system automatically backtracks and analyzes which characteristic parameters' abnormal scores contributed the most to the surge in CAI. Combined with sensor location information, it initially locates the battery module, cluster, or subsystem (such as the cooling system) that may have failed, greatly shortening the on-site troubleshooting time.
[0051] The system is configured with a feedback loop. The confirmation result after each warning (whether it is a real fault or a false alarm) will be used as a new sample to incrementally update the dynamic baseline model and fusion model, enabling the system to adapt to battery aging, environmental changes and new fault modes, and achieve self-evolution.
[0052] A comprehensive fault prediction system for energy storage systems based on multi-source information fusion and composite anomaly analysis, used to execute the aforementioned comprehensive fault prediction method for energy storage systems based on multi-source information fusion and composite anomaly analysis, includes: The data processing module is configured to acquire basic parameters and historical operating data from multiple sources, preprocess the historical operating data and basic parameters for each parameter, establish a baseline model based on the historical operating data, and obtain the anomaly score for each parameter based on the current basic parameters through the baseline model. The prediction module is configured to establish a multi-source information fusion model, output the comprehensive anomaly probability value (CAI) at the current moment based on the anomaly score of each parameter through the multi-source information fusion model; sort all the currently collected comprehensive anomaly probability values according to the time sequence to obtain the CAI curve, and issue different levels of early warning based on the comprehensive anomaly probability value and growth trend of the CAI curve.
[0053] 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.
[0054] The present invention has the following advantages: Very early warning: By capturing complex anomalies, it can issue warnings hours or even days before serious failures such as thermal runaway occur, seizing the golden time for handling.
[0055] High accuracy: The fusion of multi-source information effectively cross-verifies abnormal signals, significantly reducing false alarms and missed alarms caused by fluctuations in a single parameter.
[0056] High adaptability: The dynamic baseline model can automatically adapt to different workloads and environmental conditions, making it highly universal.
[0057] Explainability and operability: The fault tracing function points out the direction of troubleshooting, improves operation and maintenance efficiency, and lowers the technical threshold.
[0058] 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 computer-readable storage medium. This computer software product, stored in a storage medium, 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 of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0059] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for comprehensive fault prediction of energy storage system based on multi-source information fusion and composite anomaly analysis, characterized in that, The method comprises the following steps: obtaining basic parameters and historical operation data of multi-source information, preprocessing the historical operation data and the basic parameters of each parameter; establishing a baseline model based on the historical operation data, and obtaining an abnormal score of each parameter through the baseline model based on the current basic parameters; establishing a multi-source information fusion model, and outputting a comprehensive abnormal probability value CAI at the current time through the multi-source information fusion model based on the abnormal score of each parameter; sorting all the comprehensive abnormal probability values collected at the current time based on the time sequence to obtain a CAI curve, and issuing different levels of early warnings based on the comprehensive abnormal probability value and the growth trend of the CAI curve. 2.The energy storage system comprehensive fault prediction method based on multi-source information fusion and composite anomaly analysis according to claim 1, characterized in that, The preprocessing of the historical operation data and the basic parameters of each parameter comprises the following steps: performing missing value and abnormal value elimination on the historical operation data and the basic parameters, and supplementing the eliminated missing values and abnormal values by using the average value of adjacent data; normalizing the historical operation data and the basic parameters of each parameter, and outputting the normalized historical operation data and the basic parameters; In the formula, is the normalized data of the i th moment of the j th parameter, is the data of the i th moment of the j th parameter, is the minimum value of the current category parameter, is the maximum value of the current category parameter; performing time stamping on the normalized historical operation data and the basic parameters, and sorting according to the time stamp. 3.The energy storage system comprehensive fault prediction method based on multi-source information fusion and composite anomaly analysis according to claim 2, characterized in that, The establishment of the baseline model based on the historical operation data comprises the following steps: obtaining the historical operation data of each parameter, the maximum value and the minimum value at each time, establishing a confidence interval, and the confidence interval is the data range in the time sequence; connecting a plurality of maximum values to form the upper limit of the confidence interval in the time sequence, and connecting a plurality of minimum values to form the lower limit of the confidence interval in the time sequence; the data within the confidence interval is the normal fluctuation range, and the upper limit and the lower limit of the confidence interval are the health baseline; connecting the average values of a plurality of maximum values and minimum values to form an expected value curve.
4. The energy storage system comprehensive fault prediction method based on multi-source information fusion and composite anomaly analysis according to claim 3, characterized in that, The abnormal score of each parameter obtained through the baseline model based on the current basic parameters comprises the following steps: calculating the deviation of the time stamp of the basic parameters from the expected value curve of the same type of parameter to obtain an abnormal score. wherein, is the abnormal score of the jth basic parameter at the ith time point, is the total number of time series, is the data of the jth basic parameter at the ith time point, is the data of the expected value curve at the ith time point.
5. The energy storage system comprehensive fault prediction method based on multi-source information fusion and composite anomaly analysis according to claim 4, characterized in that, The establishment of the multi-source information fusion model comprises the following steps: In the formula, is the comprehensive abnormal probability value of the current energy storage system at the i-th moment, is the calculation weight of the j-th basic parameter, is the type of the basic parameter; connecting all the comprehensive abnormal probability values at all times to obtain a CAI curve.
6. The energy storage system comprehensive fault prediction method based on multi-source information fusion and composite anomaly analysis according to claim 5, characterized in that, The different levels of early warnings issued based on the comprehensive abnormal probability value and the growth trend of the CAI curve comprise the following steps: The early warning levels include a first-level early warning, a second-level early warning and a third-level early warning, and a first-level threshold value, a second-level threshold value and a third-level threshold value are set respectively; when the comprehensive abnormal probability value at the i-th time is greater than or equal to the first-level threshold value and less than the second-level threshold value, a first-level early warning is issued, and when the comprehensive abnormal probability value at the i-th time is less than the first-level threshold value, no early warning is issued; when the comprehensive abnormal probability value at the i-th time is greater than or equal to the second-level threshold value and less than the third-level threshold value, a second-level early warning is issued; when the comprehensive abnormal probability value at the i-th time is greater than or equal to the third-level threshold value, a third-level early warning is issued.
7. The energy storage system comprehensive fault prediction method based on multi-source information fusion and composite anomaly analysis according to claim 4, characterized in that, The method comprises the following steps: The method comprises the following steps: In the formula, is an over-limit rate, is the number of time nodes at which the base data exceeds the upper limit or the lower limit. When , is the judgment threshold value of the transition model, no change is made, when , r comprehensive abnormal probability values after the current time node on the CAI curve are obtained, whether the growth rate of the current comprehensive abnormal probability value rises is judged based on the r comprehensive abnormal probability values, if it rises, the current warning level is modified to the upper level of the warning level and then output, if it does not rise, no change is made.
8. A power storage system based on multi-source information fusion and composite anomaly analysis, characterized in that, The data processing module is configured to acquire basic parameters and historical operation data of multi-source information, and pre-process the historical operation data and the basic parameters of each parameter; a baseline model is established based on the historical operation data, and an abnormal score of each parameter is obtained through the baseline model based on the current basic parameters; The prediction module is configured to establish a multi-source information fusion model, output a comprehensive abnormal probability value CAI at a current time through the multi-source information fusion model based on the abnormal score of each parameter; sort all the comprehensive abnormal probability values currently collected based on a time sequence to obtain a CAI curve, and issue different levels of early warnings based on the comprehensive abnormal probability values and growth trends of the CAI curve.
9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the method for comprehensive fault prediction of the energy storage system based on multi-source information fusion and compound anomaly analysis in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores the computer program, and the computer program is executed by the processor to implement the method for comprehensive fault prediction of the energy storage system based on multi-source information fusion and compound anomaly analysis in any one of claims 1-7.