Big data based multi-level battery maintenance method and system

Abstract

The application relates to battery technology, in particular to a battery multistage maintenance method and system based on big data. The method comprises the following steps: collecting real-time operation data of a battery unit layer and a module layer, and performing pretreatment to obtain an operation parameter set; respectively performing fluctuation characteristic analysis on temperature and current load of the operation parameter set, and combining historical data to calculate corresponding state difference indexes; in the case that the state difference indexes exceed preset boundary offset threshold values, performing smoothing processing on the operation parameter set, and performing attenuation stage characteristic analysis to determine a corresponding attenuation risk level; according to the attenuation risk level, determining a corresponding control strategy, and generating a corresponding boundary correction value according to a preset amplitude limit condition; and according to the boundary correction value, respectively implementing real-time current limitation on the battery unit layer and the module layer, and continuously monitoring the operation states of the battery unit layer and the module layer. The technical problems that state differences of multistage batteries are difficult to be dynamically identified and safe boundaries are self-adaptively adjusted are solved.

Description

Technical Field

[0001] This invention belongs to the field of battery technology, specifically a multi-level battery maintenance method and system based on big data. Background Technology

[0002] In the field of modern energy management, the safety and operational efficiency of battery systems are directly related to the promotion and application of new energy technologies. Especially in scenarios such as electric vehicles and energy storage power stations, the battery, as a core component, has a decisive impact on the stability and lifespan of the overall system. With the increasing complexity of application environments, the operating performance of batteries under different operating conditions exhibits diverse characteristics. Traditional management methods based on a single dimension or fixed thresholds are no longer sufficient to meet the needs of refined maintenance.

[0003] Existing battery management technologies typically focus on monitoring single-level or local parameters, lacking the ability to collaboratively analyze multi-level structures such as battery cell and module layers, making it difficult to comprehensively reflect the overall operating status of the battery system. Furthermore, during long-term operation, battery performance is affected by a combination of factors such as temperature and current load, exhibiting dynamic changes. Existing solutions lag in state identification and strategy adjustment, resulting in the inability to take timely and effective control measures when performance degradation or anomalies occur, thus impacting system safety and operational efficiency.

[0004] Furthermore, determining battery safety boundaries is highly dynamic and varies across different levels. Different levels of battery cells exhibit different state characteristics during operation, and their safety thresholds continuously change with environmental conditions and aging. Without precise characterization and dynamic assessment of the state differences at each level, it is difficult to achieve reasonable setting and real-time adjustment of safety boundaries. Summary of the Invention

[0005] To address the above problems, this invention provides a multi-level battery maintenance method and system based on big data, which solves the problem of difficulty in dynamically identifying differences in battery state at multiple levels and adaptively adjusting safety boundaries.

[0006] To achieve the above objectives, in a first aspect, this application provides a multi-level battery maintenance method based on big data, comprising: collecting real-time operating data of the battery cell layer and the module layer, and preprocessing the real-time operating data to obtain a set of operating parameters, wherein the real-time operating data includes temperature, current load, and voltage; calling a state prediction model to perform fluctuation characteristic analysis on the temperature and current load of the operating parameter set respectively, and calculating the state difference index corresponding to the battery cell layer and the module layer in combination with historical data; when the state difference index exceeds a preset boundary offset threshold, smoothing the operating parameter set to obtain a parameter feature set, and calling a feature extraction model to perform attenuation stage feature analysis on the parameter feature set to determine the attenuation risk level corresponding to the battery cell layer and the module layer; determining the corresponding control strategy from a protection measure library according to the attenuation risk level, and generating boundary correction values ​​corresponding to the battery cell layer and the module layer according to preset amplitude limit conditions; determining the safety boundary value corresponding to the battery cell layer and the module layer according to the boundary correction value, and implementing real-time current limiting on the battery cell layer and the module layer respectively according to the safety boundary value, while continuously monitoring their operating status.

[0007] Secondly, this application provides a multi-level battery maintenance system based on big data, comprising: a preprocessing module for collecting real-time operating data of the battery cell layer and the module layer, and preprocessing the real-time operating data to obtain a set of operating parameters, wherein the real-time operating data includes temperature, current load, and voltage; a first analysis module for calling a state prediction model to perform fluctuation characteristic analysis on the temperature and current load of the set of operating parameters, and calculating the state difference index corresponding to the battery cell layer and the module layer in combination with historical data; and a second analysis module for further processing the set of operating parameters when the state difference index exceeds a preset boundary offset threshold. A smoothing process is performed to obtain a parameter feature set, and a feature extraction model is called to perform attenuation stage feature analysis on the parameter feature set to determine the attenuation risk level corresponding to the battery cell layer and the module layer. A generation module is used to determine the corresponding control strategy from the protection measure library based on the attenuation risk level, and generate the boundary correction value corresponding to the battery cell layer and the module layer according to the preset amplitude limit conditions. A control module is used to determine the safety boundary value corresponding to the battery cell layer and the module layer based on the boundary correction value, and implement real-time current limiting on the battery cell layer and the module layer respectively according to the safety boundary value, while continuously monitoring their operating status.

[0008] As an optional example, the preprocessing module includes: a data acquisition unit for acquiring real-time temperature, real-time current load, and real-time voltage data of the battery cell layer and the module layer through a sensor network to construct the real-time operating data; a first processing unit for performing outlier detection and removal processing on the real-time operating data to obtain preliminary cleaned data; a first calculation unit for calculating noise estimates on the preliminary cleaned data and recording the corresponding data accuracy information; a second processing unit for performing secondary filtering processing on the target data when the accuracy of the target data is lower than a preset accuracy threshold to obtain secondary cleaned data, wherein the target data is any one data point in the preliminary cleaned data; and an integration unit for integrating and hierarchically aggregating the parameters of the secondary cleaned data to form the set of operating parameters.

[0009] As an optional example, the first analysis module includes: a splitting unit for splitting the temperature data and current load data in the above-mentioned set of operating parameters to construct temperature data sequences and current load data sequences respectively; a first extraction unit for inputting the temperature data sequences and current load data sequences into the above-mentioned state prediction model so that the state prediction model extracts the corresponding fluctuation period features and fluctuation amplitude features to obtain fluctuation feature description results; a third processing unit for obtaining historical data segments corresponding to the fluctuation feature description results from the above-mentioned historical data, and performing weighted processing on the historical data segments according to preset weight processing rules to obtain temperature fluctuation reference values ​​and current load reference values ​​corresponding to the above-mentioned battery cell layer and the above-mentioned module layer; a second calculation unit for calculating temperature deviation values ​​and current load deviation values ​​corresponding to the above-mentioned battery cell layer and the above-mentioned module layer respectively based on the above-mentioned temperature fluctuation reference values ​​and the above-mentioned current load reference values; and a third calculation unit for calculating state difference indicators corresponding to the above-mentioned battery cell layer and the above-mentioned module layer at different levels according to preset deviation thresholds, the above-mentioned temperature deviation values, and the above-mentioned current load deviation values.

[0010] As an optional example, the second analysis module includes: a first determining unit for determining the range of abnormal parameters in the set of operating parameters; a fourth processing unit for performing sliding window smoothing on the set of operating parameters using a preset time window based on the range of abnormal parameters to obtain preliminary smoothed data; a fifth processing unit for performing weighted fusion processing on newly acquired real-time operating data and the preliminary smoothed data according to a preset measurement update frequency to obtain fused data; and a construction unit for constructing the parameter feature set based on the fused data.

[0011] As an optional example, the second analysis module includes: a second extraction unit, used to call the feature extraction model to perform feature extraction processing on the parameter feature set to obtain attenuation feature parameters, wherein the attenuation feature parameters include voltage features, internal resistance features, and capacity features; a fourth calculation unit, used to calculate the attenuation rate corresponding to the battery cell layer and the module layer respectively based on the attenuation feature parameters; a fifth calculation unit, used to acquire operating environment data and calculate the environmental coefficient based on the operating environment data; a sixth calculation unit, used to call the comprehensive scoring model and combine the environmental coefficient to score the attenuation rate respectively to obtain the evaluation score corresponding to the battery cell layer and the module layer; and a second determination unit, used to determine the attenuation risk level corresponding to the battery cell layer and the module layer respectively based on the relationship between the evaluation score and the preset grading threshold.

[0012] As an optional example, the generation module includes: a first acquisition unit, configured to acquire candidate control strategy sets corresponding to the battery cell layer and the module layer respectively from the protection measure library according to the attenuation risk level; a screening unit, configured to screen the candidate control strategy sets respectively based on preset strategy matching rules to determine the target control strategies corresponding to the battery cell layer and the module layer; and a generation unit, configured to extract the corresponding parameter adjustment rules respectively according to the target control strategies, and constrain the adjustment range in combination with the preset amplitude limit conditions to generate boundary correction values ​​corresponding to the battery cell layer and the module layer.

[0013] As an optional example, the above device further includes: a data acquisition module, configured to perform parameter adjustment based on the boundary correction value before determining the safety boundary values ​​corresponding to the battery cell layer and the module layer based on the boundary correction value, and to acquire operational feedback data within a preset feedback verification period; a calculation module, configured to calculate an anomaly rate based on the operational feedback data; a first adjustment module, configured to adjust the boundary correction value by attenuation based on the anomaly rate when the anomaly rate exceeds a preset anomaly rate threshold, to obtain an adjusted boundary correction value; a matching module, configured to acquire historical operational case data and match the operational feedback data with the historical operational case data to obtain a historical stability rate corresponding to the adjusted boundary correction value; a scoring module, configured to call an applicability assessment model to score the applicability of the adjusted boundary correction value based on the historical stability rate, and to confirm the adjusted boundary correction value when the applicability score exceeds a preset scoring threshold; and a second adjustment module, configured to further attenuate the adjusted boundary correction value when the applicability score does not exceed the preset scoring threshold, until the corresponding applicability score exceeds the preset scoring threshold.

[0014] As an optional example, the control module includes: a second acquisition unit, used to acquire the initial safety boundary values ​​corresponding to the battery cell layer and the module layer; and a seventh calculation unit, used to perform superposition calculation on the boundary correction value and the corresponding initial safety boundary value to obtain the safety boundary values ​​corresponding to the battery cell layer and the module layer.

[0015] As an optional example, the control module includes: an eighth calculation unit, used to continuously collect operating status feedback data and calculate the degree of deviation based on the operating status feedback data; a sixth processing unit, used to perform outlier removal and smoothing processing on the operating status feedback data when the degree of deviation is lower than a preset deviation threshold, to obtain a cleaned parameter dataset; and a ninth calculation unit, used to recalculate the state difference index and the attenuation risk level based on the cleaned parameter dataset, so as to adjust the boundary correction value.

[0016] The technical solutions provided in this application have the following advantages compared with the prior art: This application, based on industrial big data technology, collects temperature, current load, and voltage data from battery cell and module layers. It then analyzes fluctuation characteristics using big data preprocessing and state prediction models, calculates state difference indicators based on historical data, and analyzes attenuation characteristics through smoothing and feature extraction models when the state difference indicators exceed preset boundary offset thresholds to obtain attenuation risk levels. Furthermore, it matches control strategies to risk levels and generates boundary correction values ​​to determine safety boundaries and implement current limits. Simultaneously, it achieves dynamic optimization and adjustment through continuous monitoring, thereby realizing accurate identification and adaptive control of multi-level battery states. This improves system safety, stability, and operating efficiency, ultimately solving the technical problem of dynamically identifying multi-level battery state differences and adaptively adjusting safety boundaries. Attached Figure Description

[0017] Figure 1 This is a flowchart of an optional big data-based multi-level battery maintenance method according to an embodiment of this application.

[0018] Figure 2 This is a schematic diagram of an optional big data-based multi-level battery maintenance system according to an embodiment of this application. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solution, the present invention will be described in detail below with reference to embodiments. The description in this part is only exemplary and explanatory, and should not be used to limit the scope of protection of the present invention in any way.

[0020] According to a first aspect of the embodiments of this application, a multi-level battery maintenance method based on big data is provided, optionally, as follows: Figure 1 As shown, the above method includes: S102 collects real-time operating data from the battery cell layer and module layer, and preprocesses the real-time operating data to obtain a set of operating parameters, including temperature, current load and voltage. S104, call the state prediction model to perform fluctuation characteristic analysis on the temperature and current load of the operating parameter set, and calculate the state difference index corresponding to the battery cell layer and module layer in combination with historical data. S106, when the state difference index exceeds the preset boundary offset threshold, the set of operating parameters is smoothed to obtain the parameter feature set, and the feature extraction model is called to perform attenuation stage feature analysis on the parameter feature set to determine the attenuation risk level corresponding to the battery cell layer and the module layer. S108, based on the attenuation risk level, determines the corresponding control strategy from the protection measure library, and generates the boundary correction values ​​corresponding to the battery cell layer and module layer according to the preset amplitude limit conditions; S110: Based on the boundary correction value, determine the corresponding safety boundary value for the battery cell layer and the module layer, and implement real-time current limiting for the battery cell layer and the module layer respectively based on the safety boundary value, while continuously monitoring their operating status.

[0021] Optionally, in this embodiment, firstly, operational data of the battery cell layer and module layer are collected in real time through a sensor network deployed in the battery system. The operational data includes key parameters such as temperature, current load, and voltage. The collected real-time operational data is preprocessed, including time alignment, outlier removal, noise filtering, and data standardization, thereby obtaining a structured set of operational parameters.

[0022] Based on this, a state prediction model is invoked to analyze the fluctuation characteristics of temperature and current load in the set of operating parameters. The state prediction model can be a time-series based prediction model, such as an Autoregressive Moving Average (ARIMA) model, a Long Short-Term Memory (LSTM) network, or other sequence modeling methods, used to extract feature information such as fluctuation period and amplitude. By extracting feature information such as fluctuation period and amplitude, a feature description of the current operating state is constructed. Simultaneously, combined with historical operating data, the current features are matched and weighted to calculate the state difference index corresponding to the battery cell layer and module layer, used to characterize the degree of deviation of the current operating state from the historical stable state. When the state difference index exceeds a preset boundary offset threshold, it indicates that the battery system has potential anomalies or performance degradation trends. The boundary offset threshold is a pre-set judgment standard used to distinguish between normal fluctuations and abnormal offsets.

[0023] To address the above issues, the set of operating parameters is further processed. Methods such as sliding windows are used to smooth the data and eliminate short-term fluctuations, resulting in a parameter feature set. Subsequently, a feature extraction model is invoked to perform attenuation stage feature analysis on the parameter feature set, extracting attenuation characteristic parameters from multiple dimensions such as voltage, internal resistance, and capacity. The feature extraction model can be a statistical analysis model or a machine learning model (such as principal component analysis (PCA), clustering model, or regression model) to extract key attenuation parameters such as voltage, internal resistance, and capacity features. The attenuation rate of the battery cell layer and module layer is then calculated based on these feature parameters. Simultaneously, operating environment data (such as ambient temperature and humidity) is introduced, and environmental coefficients are calculated using preset formulas, for example, by quantifying different environmental factors using a weighted linear combination method. A comprehensive scoring model is then used to weightedly evaluate the attenuation rate, obtaining a corresponding evaluation score. This comprehensive scoring model can be a weighted summation model or a regression model, with weight parameters trained based on historical data or set empirically. By comparing the evaluation score with preset grading thresholds, the battery cell layer and module layer are divided into different attenuation risk levels (such as low risk, medium risk, and high risk) to reflect the safety level of the current operating state.

[0024] Based on the degradation risk level, corresponding control strategies are selected from a pre-built protection measure library. This library can contain various operational control strategies, such as current limiting strategies, temperature control strategies, and balancing strategies. Combined with preset amplitude limits (such as maximum adjustment amplitude or rate of change constraints), the parameter adjustment range in the control strategies is constrained, thereby generating boundary correction values ​​for the battery cell layer and module layer.

[0025] Furthermore, the boundary correction value and the initial safety boundary value are superimposed and verified in conjunction with preset safety constraints (such as the maximum allowable current range and temperature safety range) to determine the final safety boundary value. Based on the safety boundary value, real-time current limiting control is implemented on both the battery cell layer and the module layer to reduce operational risks and delay performance degradation.

[0026] During system operation, the battery's operating status is continuously monitored by cyclically collecting operational feedback data and calculating the degree of deviation. When the detected deviation decreases to a preset range, outlier values ​​are removed and smoothed from the feedback data, and the state difference index and attenuation risk level are recalculated to dynamically adjust the boundary correction value, thereby achieving closed-loop optimization control of the battery's operating status.

[0027] The above method enables dynamic sensing and precise assessment of the operating status in multi-level battery structures. Based on the assessment results, control strategies and safety boundary values ​​are adaptively adjusted, significantly improving the safety and operational stability of the battery system. Compared to traditional methods, this method fully integrates industrial big data technology to efficiently process and deeply model and analyze massive, multi-source, and heterogeneous operating data, effectively improving the accuracy of status identification and the timeliness of response. Simultaneously, through a robust data acquisition and control mechanism, real-time acquisition and efficient management of key operating parameters are achieved, effectively reducing the operational risks of the battery system, extending battery life, and improving the overall energy system's operational efficiency.

[0028] As an optional example, real-time operating data from the battery cell layer and module layer is collected, and the real-time operating data is preprocessed to obtain a set of operating parameters, including: Real-time temperature, real-time current load, and real-time voltage data of the battery cell layer and module layer are collected through a sensor network to construct real-time operating data. Outlier detection and removal are performed on real-time running data to obtain preliminary cleaned data; Noise estimates were calculated from the preliminary cleaning data, and the corresponding data accuracy information was recorded. If the accuracy of the target data is lower than the preset accuracy threshold, the target data is subjected to secondary filtering to obtain secondary cleaned data. The target data is any one of the data in the initial cleaned data. The secondary cleaning data is integrated and hierarchically categorized to form a set of operating parameters.

[0029] Optionally, in this embodiment, a sensor network deployed in the battery system is used to collect key operating parameters of the battery cell layer and module layer in real time. These operating parameters include, but are not limited to, real-time temperature, real-time current load, and real-time voltage data. Based on this data, an initial set of real-time operating data is constructed. To improve data quality, outlier detection and removal are first performed on the real-time operating data. Outlier detection can be based on statistical methods or preset engineering thresholds to remove abnormal data caused by sensor errors or communication interference, thereby obtaining preliminary cleaned data.

[0030] Based on this, noise estimates are calculated on the initially cleaned data. By analyzing the variance or fluctuation of the data, the noise level is assessed, and the corresponding data accuracy information is recorded simultaneously to reflect the data's reliability. When some data is found to have accuracy below a preset accuracy threshold, it undergoes further processing. A secondary filtering mechanism is used to screen and correct the data, removing low-reliability data to obtain the second-cleaned data. Secondary filtering can be combined with historical data distribution or multi-dimensional parameter constraints to ensure that the retained data meets accuracy requirements.

[0031] Subsequently, the secondary cleaning data underwent parameter integration and hierarchical aggregation. Data from different sources and dimensions were structured and organized according to the battery cell layer and module layer. Combined with a pre-defined data structure model, various parameters were uniformly managed, ultimately forming a standardized set of operating parameters. This preprocessing process improved data accuracy and consistency, thereby effectively enhancing the overall reliability and analytical precision of the multi-level battery maintenance method.

[0032] As an optional example, the state prediction model is invoked to analyze the fluctuation characteristics of temperature and current loads of the operating parameter set, and the state difference indicators corresponding to the battery cell layer and module layer are calculated by combining historical data, including: The temperature data and current load data in the set of operating parameters are split and processed to construct temperature data sequences and current load data sequences respectively; Temperature data sequences and current load data sequences are input into the state prediction model so that the state prediction model can extract the corresponding fluctuation period characteristics and fluctuation amplitude characteristics to obtain the fluctuation characteristic description results. Historical data segments corresponding to the fluctuation characteristic description results are obtained from historical data, and the historical data segments are weighted according to the preset weight processing rules to obtain the temperature fluctuation reference value and current load reference value corresponding to the battery cell layer and module layer. Based on the temperature fluctuation reference value and the current load reference value, calculate the temperature deviation value and current load deviation value corresponding to the battery cell layer and the module layer respectively; Based on the preset deviation threshold, temperature deviation value, and current load deviation value, the state difference indicators corresponding to the battery cell layer and module layer are calculated separately at each level.

[0033] Optionally, in this embodiment, after obtaining the set of operating parameters, the temperature data and current load data are first split and processed to construct temperature data sequences and current load data sequences respectively, so as to ensure that different types of parameters can be independently modeled and analyzed. Then, the temperature data sequences and current load data sequences are input into the state prediction model to perform time-series fluctuation characteristic analysis, extracting the corresponding fluctuation period characteristics and fluctuation amplitude characteristics to obtain fluctuation characteristic description results. For example, for the temperature data of a certain battery cell layer, the model can identify the daily periodic fluctuation characteristics and predict that it has a peak change trend during the midday period; for the current load data of the module layer, it can identify the fluctuation amplitude of the load changing with the working period.

[0034] After obtaining the fluctuation characteristic description results, historical data segments matching the current fluctuation characteristics are extracted from historical operating data. Data from different time periods are then weighted according to preset weighting rules; for example, recent data is assigned higher weights and older data lower weights, thus obtaining more timely temperature fluctuation reference values ​​and current load reference values. It should be noted that the reference values ​​for the battery cell layer and module layer are calculated independently due to their different structures and operating characteristics to ensure the accuracy of the evaluation results.

[0035] Furthermore, based on the temperature fluctuation reference value and the current load reference value, the currently collected data is compared with the reference values ​​to calculate the temperature deviation value and current load deviation value of the battery cell layer and the module layer in different time periods. For example, the temperature deviation of a certain battery cell layer is 1.8℃ and the current deviation is 8.2%, while the temperature deviation of a certain module layer is 3.2℃ and the current deviation is 12.5%.

[0036] Finally, based on preset deviation thresholds (such as temperature deviation threshold ±2.5℃ and current deviation threshold ±10%), and combining the temperature deviation value and current load deviation value, the degree of deviation is stratified, and the corresponding state difference index is calculated for each. For example, when all deviations are within the threshold range, it can be determined as a low-difference state, corresponding to a smaller state difference index; when the deviation partially or completely exceeds the threshold, it is determined as a medium or high-difference state, corresponding to a larger state difference index. Through the above method, the differences in the operating state of the battery cell layer and the module layer are quantitatively expressed, providing a basis for subsequent risk assessment and control strategy formulation.

[0037] As an optional example, smoothing the set of runtime parameters yields a set of parameter features including: Determine the range of abnormal parameters in the set of operating parameters; Based on the range of abnormal parameters, a sliding window smoothing process is applied to the set of operating parameters using a preset time window to obtain preliminary smoothed data. Based on the preset measurement update frequency, the newly collected real-time operating data and the preliminary smoothed data are weighted and fused to obtain fused data; Construct a set of parameter features based on the fused data.

[0038] Optionally, in this embodiment, anomaly detection analysis is first performed on the collected set of operating parameters to determine the range of abnormal parameters. Specifically, parameter values ​​that significantly deviate from the normal range can be identified based on preset offset thresholds or statistical characteristics (such as mean and standard deviation), thereby providing constraint boundaries for subsequent processing. For example, when the state difference index is 8.5 at a certain moment, and the preset boundary offset threshold is 5.0, it can be determined that there is abnormal fluctuation in the current operating parameters, and a smoothing process is required.

[0039] After determining the range of abnormal parameters, a sliding window smoothing process is applied to the set of operating parameters using a preset time window to obtain preliminary smoothed data. Specifically, a moving average algorithm can be used to calculate the mean of continuous sampling points within a set time window (e.g., 5 seconds) to reduce the impact of instantaneous changes on the overall trend. For example, after processing the original data such as 3000 rpm, 75.3℃, and 12.7Hz vibration frequency, a preliminary smoothed result of 2980 rpm, 74.8℃, and 12.5Hz vibration frequency can be obtained, thereby improving data stability.

[0040] Furthermore, based on a preset measurement update frequency, the newly acquired real-time operational data and the preliminary smoothed data are weighted and fused to balance the real-time performance and stability of the data. Specifically, the acquisition frequency can be dynamically adjusted according to the current operating load, for example, from 10 times per second to 8 times per second, and the weight ratio of real-time data to historical smoothed data can be set (e.g., 3:7). The fused data is obtained through weighted calculation. For example, after fusion, the rotational speed is 2985 rpm, the temperature is 75.0℃, and the vibration frequency is 12.6 Hz, ensuring that the data reflects the current trend while avoiding drastic fluctuations.

[0041] Finally, a parameter feature set is constructed based on the fused data. The fused multidimensional operating parameters are then structured and integrated to form input data for subsequent attenuation analysis and risk assessment. Simultaneously, this parameter feature set can be compared with historical operating patterns, for example, by calculating a similarity index. When the similarity falls below a preset threshold (e.g., 80%), further correction or optimization mechanisms are triggered, thereby ensuring that the obtained parameter feature set has high accuracy and reliability.

[0042] As an optional example, a feature extraction model is invoked to perform degradation stage feature analysis on the parameter feature set to determine the degradation risk level corresponding to the battery cell layer and module layer, including: The feature extraction model is invoked to perform feature extraction processing on the parameter feature set to obtain attenuation feature parameters, which include voltage features, internal resistance features and capacity features. Based on the attenuation characteristic parameters, calculate the attenuation rate corresponding to the battery cell layer and the module layer respectively; Acquire runtime environment data and calculate environment coefficients based on the runtime environment data; The degradation rate is scored and calculated by calling the comprehensive scoring model and combining the environmental coefficient, and the evaluation scores corresponding to the battery cell layer and the module layer are obtained. Based on the relationship between the evaluation score and the preset grading threshold, the degradation risk level corresponding to the battery cell layer and the module layer is determined respectively.

[0043] Optionally, in this embodiment, a pre-established feature extraction model is first invoked to perform feature extraction processing on the smoothed parameter feature set to obtain degradation characteristic parameters reflecting the battery's health status. These degradation characteristic parameters include voltage characteristics, internal resistance characteristics, and capacity characteristics. Specifically, statistical analysis and trend modeling can be performed on each parameter based on time series data, such as calculating the mean, standard deviation, and rate of change, thereby identifying potential degradation stage distribution intervals.

[0044] Based on this, the degradation rates for the battery cell layer and the module layer are calculated separately according to the degradation characteristic parameters. The degradation rate can be quantified by the magnitude of capacity decrease or the rate of internal resistance increase, for example, by comparing the percentage difference between the current capacity and the initial capacity, or by characterizing the degree of degradation by the increase in internal resistance per unit cycle. For example, if the capacity of a battery cell decreases to 99.6% of its initial value in a certain period, its degradation rate is 0.4%; while the degradation rate for the module layer is 0.35%.

[0045] Furthermore, operating environment data is acquired, and an environmental coefficient is calculated based on this data to assess the impact of the external environment on battery degradation. For example, temperature and humidity can be selected as the main influencing factors, and quantified using a weighted calculation model to obtain the environmental coefficient. When the ambient temperature is high or the humidity deviates from the normal range, the environmental coefficient increases accordingly, indicating a more significant impact on degradation.

[0046] Subsequently, a comprehensive scoring model is used to calculate the degradation rate in conjunction with the environmental coefficient, resulting in evaluation scores for the battery cell layer and the module layer. Specifically, the degradation rate and the environmental coefficient can be weighted and calculated together, for example, by setting the degradation rate weight to 0.7 and the environmental coefficient weight to 0.3, thus obtaining a comprehensive evaluation result. For instance, when the battery cell layer degradation rate is 0.4% and the environmental coefficient is 0.8, its evaluation score is 0.52; the corresponding evaluation score for the module layer is 0.485.

[0047] Finally, based on the relationship between the evaluation score and the preset grading threshold, the degradation risk level corresponding to the battery cell layer and the module layer is determined respectively. For example, an evaluation score higher than 0.5 is considered high risk, between 0.3 and 0.5 is considered medium risk, and below 0.3 is considered low risk. Therefore, in the example above, the battery cell layer is determined to be high risk, and the module layer is determined to be medium risk.

[0048] As an optional example, based on the attenuation risk level, the corresponding control strategy is determined from the protection measure library, and boundary correction values ​​for the battery cell layer and module layer are generated according to preset amplitude limits, including: Based on the degradation risk level, candidate control strategy sets corresponding to the battery cell layer and module layer are obtained from the protection measure library respectively; Based on preset strategy matching rules, the candidate control strategy set is screened to determine the target control strategy corresponding to the battery cell layer and the module layer. Based on the target control strategy, the corresponding parameter adjustment rules are extracted and the adjustment range is constrained by preset amplitude limits to generate boundary correction values ​​for the battery cell layer and module layer.

[0049] Optionally, in this embodiment, based on the degradation risk levels of the battery cell layer and the module layer, corresponding candidate control strategy sets are first obtained from a pre-built protection measure library. The protection measure library pre-stores various control schemes for different risk levels and categorizes them according to battery type, operating conditions, and hierarchical characteristics to ensure strategy adaptability. For example, when a battery cell layer is determined to be at a medium risk level, multiple candidate strategies, including "reducing the charging cutoff voltage" and "limiting the depth of discharge," can be obtained from the measure library.

[0050] Based on this, the candidate control strategy set is filtered according to preset strategy matching rules to determine the target control strategy suitable for the current operating state. The strategy matching rules can comprehensively consider factors such as risk level, equipment type, and hierarchical priority. For example, a higher weight can be set for the unit level to prioritize the strategy that has a more direct protection effect on individual cells. When there are multiple candidate strategies, the optimal solution can also be selected through a priority ranking mechanism. For example, in a medium-risk scenario, the system can prioritize "adjusting the charging upper limit voltage to 4.1V and limiting the discharge depth to 80%" as the target control strategy.

[0051] Subsequently, corresponding parameter adjustment rules are extracted according to the target control strategy, and the adjustment range is constrained by preset amplitude limits to generate boundary correction values ​​for the battery cell layer and module layer. Specifically, the voltage adjustment amplitude limit can be set to ±0.2V for the cell layer, and the discharge depth adjustment range can be set to ±10% for the module layer. Under these constraints, the target strategy is quantitatively calculated. For example, if the current charging upper limit of the cell layer is 4.25V, a boundary correction value of -0.15V can be generated to adjust it to 4.1V; if the current discharge depth of the module layer is 88%, a correction value of -8% is generated to limit it to 80%. Through the above method, refined parameter correction for different layers is achieved while ensuring that the adjustment amplitude is safe and controllable.

[0052] As an optional example, before determining the safety boundary values ​​corresponding to the battery cell layer and the module layer based on the boundary correction values, the above method further includes: Parameter control is implemented based on boundary correction values, and operational feedback data is collected within a preset feedback verification period; Calculate the anomaly rate based on operational feedback data; If the anomaly rate exceeds the preset anomaly rate threshold, the boundary correction value is adjusted by attenuation according to the anomaly rate to obtain the adjusted boundary correction value. Acquire historical operational case data and match operational feedback data with historical operational case data to obtain the historical stability rate corresponding to the adjusted boundary correction value; The applicability assessment model is invoked to score the applicability of the adjusted boundary correction value based on the historical stability rate, and the adjusted boundary correction value is confirmed if the applicability score exceeds the preset scoring threshold. If the applicability score does not exceed the preset score threshold, the adjusted boundary correction value is further attenuated until the corresponding applicability score exceeds the preset score threshold.

[0053] Optionally, in this embodiment, preliminary adjustments are made to the system operating parameters based on boundary correction values, and operating feedback data is continuously collected within a preset feedback verification period to reflect the actual operating effect after the current parameter adjustment. For example, the feedback verification period can be set to 24 hours, and an operating feedback dataset is formed by real-time monitoring of data such as current, temperature, and abnormal event records.

[0054] After acquiring operational feedback data, the anomaly rate is calculated based on this data to quantify the system's stability level under the current parameter configuration. For example, if the proportion of anomalies recorded within a feedback cycle to the total number of runs is 3%, the anomaly rate can be determined to be 3%. When the anomaly rate exceeds a preset anomaly rate threshold (e.g., 2%), it indicates that the current boundary correction value is over-adjusted or unsuitable, requiring optimization. In this case, the boundary correction value is attenuated based on the anomaly rate, for example, by reducing the correction magnitude through proportional reduction, thereby obtaining the adjusted boundary correction value.

[0055] Furthermore, historical operational case data is acquired, and the current operational feedback data is matched and analyzed with the historical operational case data to obtain the historical stability rate corresponding to the adjusted boundary correction value. Specifically, this can be achieved by filtering boundary configurations and their stability performance under similar historical operating conditions as a reference for the current adjustment results. For example, historical data showing a system stability rate of 98% corresponding to a certain correction magnitude can serve as a valid reference.

[0056] Subsequently, the applicability assessment model is invoked to score the adjusted boundary correction value based on historical stability rates to evaluate its feasibility in the current system. For example, a comprehensive score can be calculated by integrating historical stability rates with the current operating status and compared with a preset scoring threshold (e.g., 95%). When the applicability score exceeds the preset scoring threshold, the adjusted boundary correction value is confirmed as a valid correction value; conversely, if the applicability score does not reach the threshold, the boundary correction value is further attenuated and the above assessment process is repeated until the corresponding applicability score meets the requirements.

[0057] Through the above steps, an adaptive optimization process for boundary correction values ​​based on feedback data and historical experience is achieved, ensuring the adjustment effect while avoiding over-correction.

[0058] As an optional example, determining the safety boundary values ​​corresponding to the battery cell layer and module layer based on the boundary correction values ​​includes: Obtain the initial safety boundary values ​​corresponding to the battery cell layer and module layer; The boundary correction value is superimposed with the corresponding initial safety boundary value to obtain the safety boundary values ​​for the battery cell layer and the module layer.

[0059] Optionally, in this embodiment, the initial safety boundary values ​​corresponding to the battery cell layer and the module layer are first obtained. The initial safety boundary values ​​are usually preset based on battery design parameters, historical operating experience, and safety specifications, such as the upper limit voltage and maximum allowable current of the battery cell layer, or the overall depth of discharge and safe voltage range of the module layer, which are used as basic constraints for system operation.

[0060] Based on this, the boundary correction value is superimposed on the corresponding initial safety boundary value to obtain the updated safety boundary value. Specifically, the boundary correction value reflects the dynamic adjustment range of the original safety boundary under the current operating state and risk assessment results, and is used to correct the initial boundary through addition or subtraction. For example, if the initial safety boundary value of a battery cell layer is 50.0 (which can be represented as a current limit value or equivalent control parameter), and the corresponding generated boundary correction value is +5.0, then the new safety boundary value is 55.0 obtained through superposition. Similarly, if the initial safety boundary value of a module layer is 80%, and the corresponding correction value is -8%, then the updated safety boundary value is 72%.

[0061] Furthermore, to ensure the rationality and feasibility of the safety boundary values, preset boundary constraints can be used to verify the results during the superposition calculation process. For example, the safety boundary values ​​can be limited to not exceeding the maximum allowable limit of the equipment or falling below the minimum operating threshold, thereby avoiding the risks caused by over-adjustment. Through the above methods, the static initial safety boundary and the dynamically generated boundary correction values ​​are merged to form a dynamic safety boundary configuration adapted to the current operating state.

[0062] As an optional example, continuous monitoring of its operational status includes: Continuously collect operational status feedback data and calculate the degree of deviation based on the operational status feedback data; If the offset is less than the preset offset threshold, outlier removal and smoothing are performed on the running status feedback data to obtain a cleaned parameter dataset. Based on the cleaned parameter dataset, the state difference index and attenuation risk level are recalculated to adjust the boundary correction values.

[0063] Optionally, in this embodiment, the operating status feedback data of the battery cell layer and module layer is continuously collected through iterative loops, and the offset degree corresponding to the current state difference is calculated based on the operating status feedback data. When the offset degree is lower than a preset offset threshold, it indicates that the system as a whole is in a relatively stable state. At this time, a data self-update mechanism is triggered to perform outlier removal and time series smoothing on the operating status feedback data to obtain a clean parameter dataset after noise suppression. Subsequently, based on the cleaned parameter dataset, the temperature data sequence and current load data sequence are reconstructed, and the state prediction model and feature extraction model are called to perform joint analysis to update the state difference index and attenuation risk level corresponding to the battery cell layer and module layer. On this basis, the current boundary correction value is adaptively fine-tuned according to the updated state difference index and attenuation risk level, thereby realizing continuous calibration and optimization of the system operating status.

[0064] For example, in practical applications, when the detected offset decreases from 6.2 to 3.1 and falls below the threshold of 5.0, a data update process is triggered. Newly acquired temperature, current, and voltage data undergo a 3x outlier removal and sliding window smoothing process to obtain a stable data sequence. Based on this data, the state difference index is recalculated, for example, revised from 4.8 to 3.6. Simultaneously, the degradation risk level is updated from "medium risk" to "low risk," and the boundary correction value is slightly adjusted for convergence, for example, from 5.0 to 4.7, to avoid performance loss due to over-protection. Through this process, adaptive optimization updates are achieved during the system's stable phase, ensuring a dynamic balance between battery safety and efficiency, and forming a closed-loop control mechanism from monitoring and evaluation to adjustment.

[0065] It should be noted that, in this document, the terms "comprising," "including," and any other variations are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Specific examples have been used in this document to illustrate the principles and implementation methods of the present invention. These examples are merely for the purpose of helping to understand the method and core ideas of the present invention. The above descriptions are only preferred embodiments of the present invention. It should be pointed out that, due to the limitations of written expression and the objective existence of infinite specific structures, those skilled in the art can make several improvements, modifications, or variations without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, variations, or combinations, or the direct application of the concept and technical solution of the present invention to other situations without modification, should all be considered within the scope of protection of the present invention.

[0066] According to another aspect of the embodiments of this application, a multi-level battery maintenance system based on big data is also provided, such as... Figure 2 As shown, it includes: The preprocessing module 202 is used to collect real-time operating data of the battery cell layer and module layer, and preprocess the real-time operating data to obtain a set of operating parameters, including temperature, current load and voltage. The first analysis module 204 is used to call the state prediction model to perform fluctuation characteristic analysis on the temperature and current load of the operating parameter set, and calculate the state difference index corresponding to the battery cell layer and the module layer in combination with historical data. The second analysis module 206 is used to smooth the set of operating parameters when the state difference index exceeds the preset boundary offset threshold, obtain the parameter feature set, and call the feature extraction model to perform attenuation stage feature analysis on the parameter feature set in order to determine the attenuation risk level corresponding to the battery cell layer and the module layer. The generation module 208 is used to determine the corresponding control strategy from the protection measure library according to the attenuation risk level, and generate the boundary correction values ​​corresponding to the battery cell layer and the module layer according to the preset amplitude limit conditions. The control module 210 is used to determine the safety boundary values ​​corresponding to the battery cell layer and the module layer according to the boundary correction value, and to implement real-time current limiting on the battery cell layer and the module layer respectively according to the safety boundary values, while continuously monitoring their operating status.

[0067] It should be noted that the preprocessing module 202 in this embodiment can be used to execute step S102 in this application embodiment, the first analysis module 204 in this embodiment can be used to execute step S104 in this application embodiment, the second analysis module 206 in this embodiment can be used to execute step S106 in this application embodiment, the generation module 208 in this embodiment can be used to execute step S108 in this application embodiment, and the control module 210 in this embodiment can be used to execute step S110 in this application embodiment.

[0068] As an optional example, the preprocessing module includes: The acquisition unit is used to collect real-time temperature, real-time current load, and real-time voltage data of the battery cell layer and module layer through a sensor network to construct real-time operating data. The first processing unit is used to detect and remove outliers from real-time running data to obtain preliminary cleaned data. The first calculation unit is used to calculate the noise estimate of the preliminary cleaning data and record the corresponding data accuracy information. The second processing unit is used to perform secondary filtering on the target data when the accuracy of the target data is lower than a preset accuracy threshold, to obtain secondary cleaned data, wherein the target data is any one data in the initial cleaned data. The integration unit is used to integrate and hierarchically collect secondary cleaning data to form a set of operating parameters.

[0069] As an optional example, the first analysis module includes: The splitting unit is used to split the temperature data and current load data in the set of operating parameters to construct temperature data sequences and current load data sequences respectively. The first extraction unit is used to input the temperature data sequence and the current load data sequence into the state prediction model so that the state prediction model can extract the corresponding fluctuation period characteristics and fluctuation amplitude characteristics to obtain the fluctuation characteristic description results. The third processing unit is used to obtain historical data segments corresponding to the fluctuation characteristic description results from historical data, and to perform weighted processing on the historical data segments according to the preset weight processing rules to obtain the temperature fluctuation reference value and current load reference value corresponding to the battery cell layer and the module layer. The second calculation unit is used to calculate the temperature deviation value and current load deviation value corresponding to the battery cell layer and the module layer respectively, based on the temperature fluctuation reference value and the current load reference value. The third calculation unit is used to calculate the state difference indicators corresponding to the battery cell layer and the module layer respectively, based on the preset deviation threshold, temperature deviation value and current load deviation value.

[0070] As an optional example, the second analysis module includes: The first determining unit is used to determine the range of abnormal parameters in the set of operating parameters; the fourth processing unit is used to perform sliding window smoothing on the set of operating parameters according to the range of abnormal parameters using a preset time window to obtain preliminary smoothed data. The fifth processing unit is used to perform weighted fusion processing on the newly collected real-time running data and the preliminary smoothed data according to the preset measurement update frequency to obtain fused data; The building unit is used to construct a set of parameter features based on the fused data.

[0071] As an optional example, the second analysis module includes: The second extraction unit is used to call the feature extraction model to perform feature extraction processing on the parameter feature set to obtain attenuation feature parameters, including voltage features, internal resistance features and capacity features. The fourth calculation unit is used to calculate the attenuation rate corresponding to the battery cell layer and the module layer respectively based on the attenuation characteristic parameters; The fifth calculation unit is used to acquire operating environment data and calculate environmental coefficients based on the operating environment data; The sixth calculation unit is used to call the comprehensive scoring model and combine it with the environmental coefficient to calculate the degradation rate and obtain the evaluation scores corresponding to the battery cell layer and the module layer. The second determining unit is used to determine the degradation risk level corresponding to the battery cell layer and the module layer respectively based on the relationship between the evaluation score and the preset grading threshold.

[0072] As an optional example, the generated modules include: The first acquisition unit is used to acquire the candidate control strategy sets corresponding to the battery cell layer and the module layer respectively from the protection measure library according to the attenuation risk level; The filtering unit is used to filter the candidate control strategy set based on the preset strategy matching rules to determine the target control strategy corresponding to the battery cell layer and the module layer. The generation unit is used to extract the corresponding parameter adjustment rules according to the target control strategy, and to constrain the adjustment range in combination with preset amplitude limit conditions, so as to generate the boundary correction values ​​corresponding to the battery cell layer and the module layer.

[0073] As an optional example, the above-described apparatus further includes: The data acquisition module is used to implement parameter control based on the boundary correction value before determining the safety boundary value corresponding to the battery cell layer and the module layer based on the boundary correction value, and to collect operation feedback data within the preset feedback verification period. The calculation module is used to calculate the anomaly rate based on the operation feedback data; The first adjustment module is used to attenuate the boundary correction value according to the anomaly rate when the anomaly rate exceeds the preset anomaly rate threshold, so as to obtain the adjusted boundary correction value. The matching module is used to acquire historical running case data and match the running feedback data with the historical running case data to obtain the historical stability rate corresponding to the adjusted boundary correction value. The scoring module is used to call the applicability assessment model to score the applicability of the adjusted boundary correction value based on the historical stability rate, and to confirm the adjusted boundary correction value if the applicability score exceeds the preset scoring threshold. The second adjustment module is used to further attenuate the adjusted boundary correction value if the applicability score does not exceed the preset score threshold, until the corresponding applicability score exceeds the preset score threshold.

[0074] As an optional example, the control module includes: The second acquisition unit is used to acquire the initial safety boundary values ​​corresponding to the battery cell layer and the module layer; The seventh calculation unit is used to superimpose the boundary correction value with the corresponding initial safety boundary value to obtain the safety boundary values ​​corresponding to the battery cell layer and the module layer.

[0075] As an optional example, the control module includes: The eighth calculation unit is used to continuously collect operation status feedback data and calculate the degree of offset based on the operation status feedback data; The sixth processing unit is used to remove outliers and smooth the running status feedback data when the offset is lower than the preset offset threshold, so as to obtain a cleaned parameter dataset. The ninth calculation unit is used to recalculate the state difference index and attenuation risk level based on the cleaned parameter dataset, so as to adjust the boundary correction value.

[0076] For other examples of this embodiment, please refer to the examples above, which will not be repeated here.

Claims

1. A multi-level battery maintenance method based on big data, characterized in that, include: Real-time operating data of the battery cell layer and module layer are collected and preprocessed to obtain a set of operating parameters, wherein the real-time operating data includes temperature, current load and voltage; The state prediction model is invoked to analyze the fluctuation characteristics of the temperature and current load of the set of operating parameters, and the state difference index corresponding to the battery cell layer and the module layer is calculated by combining historical data. If the state difference index exceeds the preset boundary offset threshold, the set of operating parameters is smoothed to obtain a set of parameter features, and the feature extraction model is called to perform attenuation stage feature analysis on the set of parameter features to determine the attenuation risk level corresponding to the battery cell layer and the module layer. Based on the attenuation risk level, a corresponding control strategy is determined from the protection measure library, and boundary correction values ​​corresponding to the battery cell layer and the module layer are generated according to the preset amplitude limit conditions. Based on the boundary correction value, the safety boundary values ​​corresponding to the battery cell layer and the module layer are determined, and real-time current limiting is implemented on the battery cell layer and the module layer respectively according to the safety boundary values, while continuously monitoring their operating status.

2. The method according to claim 1, characterized in that, The real-time operating data of the battery cell layer and module layer is collected, and the real-time operating data is preprocessed to obtain a set of operating parameters, including: The real-time temperature, real-time current load, and real-time voltage data of the battery cell layer and the module layer are collected through a sensor network to construct the real-time operating data. The real-time running data is subjected to outlier detection and removal to obtain preliminary cleaned data; The noise estimate of the preliminary cleaning data is calculated, and the corresponding data accuracy information is recorded. If the accuracy of the target data is lower than a preset accuracy threshold, the target data is subjected to secondary filtering to obtain secondary cleaned data, wherein the target data is any one of the data in the initial cleaned data; The secondary cleaning data is integrated and hierarchically categorized to form the set of operating parameters.

3. The method according to claim 1, characterized in that, The invoked state prediction model performs fluctuation characteristic analysis on the temperature and current load of the operating parameter set, and calculates the state difference indicators corresponding to the battery cell layer and the module layer based on historical data, including: The temperature data and current load data in the set of operating parameters are split and processed to construct temperature data sequences and current load data sequences respectively; The temperature data sequence and the current load data sequence are input into the state prediction model so that the state prediction model can extract the corresponding fluctuation period features and fluctuation amplitude features to obtain the fluctuation feature description results. Historical data segments corresponding to the fluctuation characteristic description results are obtained from the historical data, and the historical data segments are weighted according to the preset weight processing rules to obtain the temperature fluctuation reference value and current load reference value corresponding to the battery cell layer and the module layer. Based on the temperature fluctuation reference value and the current load reference value, calculate the temperature deviation value and current load deviation value corresponding to the battery cell layer and the module layer, respectively. Based on the preset deviation threshold, the temperature deviation value, and the current load deviation value, the state difference index corresponding to the battery cell layer and the module layer is calculated at each level.

4. The method according to claim 1, characterized in that, The smoothing process performed on the set of operating parameters to obtain the parameter feature set includes: Determine the range of abnormal parameters in the set of operating parameters; Based on the range of abnormal parameters, a sliding window smoothing process is applied to the set of operating parameters using a preset time window to obtain preliminary smoothed data. According to the preset measurement update frequency, the newly collected real-time running data and the preliminary smoothed data are weighted and fused to obtain fused data; The parameter feature set is constructed based on the fused data.

5. The method according to claim 1, characterized in that, The step of calling the feature extraction model to perform attenuation stage feature analysis on the parameter feature set to determine the attenuation risk level corresponding to the battery cell layer and the module layer includes: The feature extraction model is invoked to perform feature extraction processing on the parameter feature set to obtain attenuation feature parameters, wherein the attenuation feature parameters include voltage features, internal resistance features, and capacity features; Based on the attenuation characteristic parameters, the attenuation rates corresponding to the battery cell layer and the module layer are calculated respectively. Acquire runtime environment data and calculate environmental coefficients based on the runtime environment data; The degradation rate is scored and calculated by calling the comprehensive scoring model and combining the environmental coefficient to obtain the evaluation scores corresponding to the battery cell layer and the module layer. Based on the relationship between the evaluation score and the preset grading threshold, the degradation risk level corresponding to the battery cell layer and the module layer is determined respectively.

6. The method according to claim 1, characterized in that, The step of determining the corresponding control strategy from the protection measure library based on the attenuation risk level, and generating boundary correction values ​​for the battery cell layer and the module layer according to preset amplitude limits, includes: Based on the attenuation risk level, candidate control strategy sets corresponding to the battery cell layer and the module layer are obtained from the protection measure library respectively; Based on preset strategy matching rules, the candidate control strategy set is filtered to determine the target control strategy corresponding to the battery cell layer and the module layer. Based on the target control strategy, the corresponding parameter adjustment rules are extracted respectively, and the adjustment range is constrained by the preset amplitude limit conditions to generate the boundary correction values ​​corresponding to the battery cell layer and the module layer.

7. The method according to claim 1, characterized in that, Before determining the safety boundary values ​​corresponding to the battery cell layer and the module layer based on the boundary correction value, the method further includes: Based on the boundary correction value, parameter adjustment is implemented, and operational feedback data is collected within a preset feedback verification period; The anomaly rate is calculated based on the operational feedback data. If the anomaly rate exceeds a preset anomaly rate threshold, the boundary correction value is attenuated and adjusted according to the anomaly rate to obtain the adjusted boundary correction value. Acquire historical running case data and match the running feedback data with the historical running case data to obtain the historical stability rate corresponding to the adjusted boundary correction value; The applicability assessment model is invoked to score the applicability of the adjusted boundary correction value based on the historical stability rate, and the adjusted boundary correction value is confirmed if the applicability score exceeds a preset scoring threshold. If the applicability score does not exceed the preset score threshold, the adjusted boundary correction value is further attenuated until the corresponding applicability score exceeds the preset score threshold.

8. The method according to claim 1, characterized in that, The step of determining the safety boundary values ​​corresponding to the battery cell layer and the module layer based on the boundary correction value includes: Obtain the initial safety boundary values ​​corresponding to the battery cell layer and the module layer; The boundary correction value is superimposed with the corresponding initial safety boundary value to obtain the safety boundary value corresponding to the battery cell layer and the module layer.

9. The method according to claim 1, characterized in that, The continuous monitoring of its operational status includes: Continuously collect operational status feedback data and calculate the degree of deviation based on the operational status feedback data; If the degree of offset is lower than a preset offset threshold, outlier removal and smoothing are performed on the running status feedback data to obtain a cleaned parameter dataset. Based on the cleaned parameter dataset, the state difference index and the attenuation risk level are recalculated to adjust the boundary correction value.

10. A multi-level battery maintenance system based on big data, characterized in that, include: The preprocessing module is used to collect real-time operating data from the battery cell layer and the module layer, and to preprocess the real-time operating data to obtain a set of operating parameters, wherein the real-time operating data includes temperature, current load and voltage. The first analysis module is used to call the state prediction model to perform fluctuation characteristic analysis on the temperature and current load of the set of operating parameters, and to calculate the state difference index corresponding to the battery cell layer and the module layer in combination with historical data. The second analysis module is used to smooth the set of operating parameters when the state difference index exceeds the preset boundary offset threshold, to obtain a set of parameter features, and to call the feature extraction model to perform attenuation stage feature analysis on the set of parameter features in order to determine the attenuation risk level corresponding to the battery cell layer and the module layer. The generation module is used to determine the corresponding control strategy from the protection measure library according to the attenuation risk level, and generate the boundary correction values ​​corresponding to the battery cell layer and the module layer according to the preset amplitude limit conditions. The control module is used to determine the safety boundary values ​​corresponding to the battery cell layer and the module layer according to the boundary correction value, and to implement real-time current limiting on the battery cell layer and the module layer respectively according to the safety boundary values, while continuously monitoring their operating status.

Images