A method and system for predicting the remaining battery life of an electrochemical energy storage system

Abstract

This invention relates to the field of electrochemical energy storage technology and discloses a method and system for predicting the remaining lifespan of batteries in electrochemical energy storage systems. The method collects multi-dimensional operational data, such as temperature distribution, current ripple, and voltage decay curves, during the charge-discharge cycle of the target battery pack. It then generates feature weight parameters by combining real-time operating conditions with historical database matching results. Degradation feature patterns throughout the entire lifespan are extracted as a benchmark template to establish a dynamic degradation feature space. A multi-scale coupled lifespan prediction model is constructed, and an adaptive evaluation function is built by iteratively updating the model parameters and the physical field coupling parameters of the benchmark template. Cross-cycle data verification is triggered to perform operating condition compatibility checks. A feedback matrix is ​​generated based on the prediction error index and the verification results to dynamically adjust the mapping relationship between feature weights and the benchmark template. This method improves the accuracy and adaptability of battery remaining lifespan prediction and is suitable for electrochemical energy storage systems under complex operating conditions.

Description

Technical Field

[0001] This invention relates to the field of electrochemical energy storage technology, specifically to a method and system for predicting the remaining lifespan of batteries in an electrochemical energy storage system. Background Technology

[0002] In electrochemical energy storage systems, accurate prediction of battery remaining life is a crucial prerequisite for ensuring stable system operation. With the rapid development of the new energy industry, electrochemical energy storage technology is widely used in scenarios such as renewable energy grid connection and smart grid peak shaving. As the core component of energy storage systems, the battery's lifespan degradation characteristics directly affect the system's economy and safety.

[0003] Currently, most battery remaining life prediction methods rely on single-dimensional operational data, such as estimating life solely based on voltage or current changes. This fails to comprehensively reflect the battery's degradation state under complex operating conditions. In actual operation, battery packs often face issues such as dynamic charge / discharge rates and uneven temperature distribution, making prediction models based on a single data dimension prone to significant errors. Furthermore, existing models often employ fixed degradation feature templates, failing to consider the dynamic changes in battery degradation modes across different operating cycles, leading to a gradual decrease in prediction accuracy over long-term cyclic use.

[0004] Traditional prediction methods lack effective cross-cycle data verification mechanisms. When battery operating conditions change abruptly, the model struggles to quickly adapt to the new operating mode, easily leading to distorted prediction results. These issues make it difficult for existing prediction methods to meet the high accuracy and adaptability requirements of electrochemical energy storage systems for battery lifetime prediction in practical applications, thus limiting the efficient management and optimized operation of energy storage systems. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for predicting the remaining life of batteries in an electrochemical energy storage system, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a method for predicting the remaining life of an electrochemical energy storage system battery, the method comprising:

[0007] Multi-dimensional operating data of the target battery pack during the charge and discharge cycle are collected. The multi-dimensional operating data includes temperature distribution data, current ripple data and voltage decay curve. Feature weight parameters are generated based on the matching results of real-time operating conditions and historical operating database.

[0008] Degradation feature patterns are extracted throughout the entire life cycle of the target battery pack. These degradation feature patterns are generated by analyzing the capacity decay trajectory under different charge and discharge rates, and are used as a benchmark template for establishing a dynamic degradation feature space.

[0009] A multi-scale coupled lifetime prediction model is constructed, driving the model parameters to iteratively update in the direction of minimizing prediction error. Each iteration generates feature reconstruction instructions and updates the physical field coupling parameters of the benchmark template. An adaptive evaluation function for the model is constructed based on the correlation between feature weight parameters and real-time running data.

[0010] Cross-cycle data verification is triggered based on the amplitude of the feature reconstruction instruction. The cross-cycle data verification extracts a verification data packet that matches the current charge and discharge mode from the historical operation database and performs operating condition compatibility verification on the verification data packet through the electrochemical conversion rules in the benchmark template.

[0011] The prediction error index output by the model adaptive evaluation function and the verification results of the verification data package are used to generate a feedback matrix, and the mapping relationship between the feature weight parameters and the benchmark template is dynamically adjusted.

[0012] Preferably, the construction of a multi-scale coupled lifetime prediction model, driving the model parameters to iteratively update in the direction of minimizing prediction error, includes:

[0013] The degradation feature patterns are mapped to high-dimensional tensors, with each dimension corresponding to the adjustable range of physical field coupling parameters. The correlation coefficients of electrochemical parameters are calculated based on the current charge and discharge modes.

[0014] Based on the correlation coefficient, an error neighborhood table is constructed to record historical failure cases that are associated with the current degradation feature mode and their differences in physical field coupling parameters, thereby generating a candidate reconstruction instruction set.

[0015] A dual-constraint screening mechanism is used to select target reconstruction instructions from the candidate reconstruction instruction set. The constraints include the error reduction rate of the candidate reconstruction instructions in historical corrections and the parameter compatibility threshold with the electrochemical conversion rules.

[0016] The target reconstruction instruction is applied to the degenerate feature pattern to dynamically update the parameter values ​​of the corresponding dimension in the high-dimensional tensor, while adjusting the filtering threshold of the error neighborhood table to trigger the elimination of low-relevance data.

[0017] The process involves recalculating the matching degree between the updated degraded feature pattern and the verification data packet, and feeding back the change in matching degree to the adjustment of the feature weight parameters.

[0018] Preferably, the step of selecting the target reconstruction instruction from the candidate reconstruction instruction set using a dual-constraint screening mechanism includes:

[0019] The error reduction rate of each candidate reconstruction instruction within the preset time window is statistically analyzed, and the ratio of the number of historical successful corrections to the change in the priority of reconstruction instructions is normalized to the initial screening weight.

[0020] Extract the parameter constraint fields corresponding to the electrochemical conversion rules in the benchmark template, calculate the coverage ratio of the parameter dimensions corresponding to the candidate reconstruction instructions and the preset constraint fields, and map them as compatibility constraint weights;

[0021] The initial screening weights and compatibility constraint weights are dynamically superimposed, and the superposition coefficient is adjusted according to the distribution density of low-correlation data in the error neighborhood table, so that the compatibility constraint weights can obtain a higher proportion in the high-risk area of ​​prediction error.

[0022] Based on the comprehensive screening probability, the instruction selection operation is performed to verify the compatibility between the candidate reconstruction instructions and the core parameters of the baseline template. If there is a conflict, the backup reconstruction instruction queue is activated.

[0023] Preferably, the step of dynamically superimposing the initial screening weights and compatibility constraint weights includes:

[0024] Based on the comparison between the proportion of low-relevance data in the error neighborhood table and the historical maximum data volume, the error risk level is calculated through a multi-segment function.

[0025] Dynamic superposition factors are generated based on the error risk level, and a threshold segmented adjustment mechanism is adopted, combined with the historical moving average as a stability factor to control the fluctuation range of the superposition factors.

[0026] The initial screening weights are correlated with the dynamic superposition factors, and the correlation results are fused with the compatibility constraint weights.

[0027] The strength of the stability factor is adjusted based on the degree of deviation between the superposition factor and the historical moving average.

[0028] Preferably, the adaptive evaluation function for the model based on the correlation between feature weight parameters and real-time running data includes:

[0029] Based on the distribution characteristics of the feature weight parameters, the spatiotemporal correlation between the parameters of each dimension and the battery degradation rate in the real-time running data is calculated.

[0030] A dynamic correction factor is generated based on spatiotemporal correlation, which correlates the parameter change trends of high decay rate regions with temperature anomaly regions, and the scope of the correction factor is constrained by the physical field coupling parameters of the benchmark template.

[0031] By integrating feature weight parameters, dynamic correction factors, and spatiotemporal correlation, an adaptive evaluation function for the model is constructed, and a penalty term related to the deviation of the physical field coupling parameters is introduced.

[0032] Preferably, the generation of the dynamic correction factor based on spatiotemporal correlation includes:

[0033] Based on the spatiotemporal correlation and the inverse correlation between electrochemical parameter fluctuations and capacity decay gradient, the difference in parameter changes between the high decay rate region and the temperature anomaly region is calculated.

[0034] An initial correction factor is generated based on the difference in parameter changes. The parameter sensitivity in the high decay rate region is correlated with the electrochemical response delay in the temperature anomaly region. The rate of change of the correction factor is constrained by the physical field coupling parameters of the reference template.

[0035] Based on the historical correction records of the physical field coupling parameters, an attenuation control factor is applied to the initial correction factor.

[0036] Preferably, the step of generating a feedback matrix by combining the prediction error index output by the model adaptive evaluation function with the verification data packet validation result includes:

[0037] The error-compatibility coupling coefficient is calculated based on the prediction error index and the compatibility compliance rate in the verification data packet results.

[0038] An initial feedback matrix is ​​generated based on the error and compatibility coupling coefficient. The lack of compatibility in high error regions is associated with feature reconstruction instructions. The update strength of the constraint feedback matrix is ​​constrained by the historical correction record of the benchmark template.

[0039] A multidimensional feedback model is constructed by fusing the coupling coefficient between error and compatibility, the prediction error index, and the compatibility achievement rate.

[0040] Preferably, the method further includes:

[0041] Establish a quantitative model of battery pack degradation state and analyze the main and secondary feature quantities in multi-dimensional operating data;

[0042] Construct a degradation feature space topology network and encode the historical degradation patterns of different charge and discharge stages as topology nodes;

[0043] The similarity path between the current running data and the topological nodes is calculated using a dynamic time warping algorithm.

[0044] When the similarity path exceeds the preset threshold, the prediction model parameter reset mechanism is triggered.

[0045] Preferably, the step of calculating the similarity path between the current running data and the topology nodes using the dynamic time warping algorithm includes:

[0046] The current running data is segmented using a sliding window to generate multiple data fragments and their timestamp indices.

[0047] Extract the electrochemical feature vector of each data segment and calculate the dynamic bending path with respect to the feature vector of the topological node;

[0048] A similarity matrix is ​​generated based on a dynamic curved path, and the optimal matching path is located through a backtracking algorithm.

[0049] Model correction instructions are generated based on the degree of deviation between the optimal matching path and the preset threshold.

[0050] Preferably, the present invention further includes a prediction system for the remaining battery life of an electrochemical energy storage system, used to implement the prediction method for the remaining battery life of an electrochemical energy storage system as described above, the system comprising:

[0051] The data acquisition module collects multi-dimensional operating data of the target battery pack during the charge and discharge cycle, and generates feature weight parameters based on the matching results of real-time operating conditions and historical operating database.

[0052] The template construction module extracts the degradation feature patterns throughout the entire life cycle of the target battery pack and uses these degradation feature patterns as a reference template for establishing a dynamic degradation feature space.

[0053] The model iteration module constructs a multi-scale coupled lifetime prediction model, drives the model parameters to be updated iteratively in the direction of minimizing the prediction error, and constructs an adaptive evaluation function for the model based on the correlation between feature weight parameters and real-time running data.

[0054] The verification trigger module triggers cross-cycle data verification based on the amplitude of the feature reconstruction instruction, and performs operating condition compatibility verification on the verification data package through the electrochemical conversion rules in the benchmark template.

[0055] The feedback generation module generates a feedback matrix by combining the prediction error index output by the model's adaptive evaluation function with the verification results of the validation data package, and dynamically adjusts the mapping relationship between the feature weight parameters and the benchmark template.

[0056] Compared with the prior art, the beneficial effects of the present invention are:

[0057] By collecting multi-dimensional operational data, including information such as temperature distribution, current ripple, and voltage decay curves, the degradation characteristics of batteries under different operating conditions can be comprehensively captured, avoiding the one-sidedness of information caused by a single data dimension. By combining the matching results of real-time operating conditions with historical databases to generate feature weight parameters, the model can dynamically adjust the influence weight of each feature according to the actual operating state, improving its adaptability to complex operating conditions.

[0058] By extracting degradation feature patterns throughout the entire battery lifecycle as a baseline template and establishing a dynamic degradation feature space based on it, the limitations of traditional fixed templates are overcome. Analyzing the capacity decay trajectory under different charge / discharge rates enables the baseline template to reflect the degradation patterns of the battery under diverse operating conditions, providing a more realistic feature reference for subsequent predictions. Simultaneously, updating the physical field coupling parameters of the baseline template during model iteration further enhances the template's ability to track the dynamic degradation process of the battery.

[0059] The multi-scale coupled lifetime prediction model improves prediction accuracy by iteratively optimizing parameters to minimize prediction error. An adaptive evaluation function, constructed by combining feature weight parameters with real-time running data, can monitor the model's prediction performance in real time and dynamically adjust model parameters based on error conditions, ensuring stable prediction performance over long-term operation.

[0060] The introduction of a cross-cycle data validation mechanism effectively addresses the problem of insufficient model adaptability during sudden changes in operating conditions by extracting matching validation data packages from historical databases and utilizing electrochemical conversion rules in benchmark templates for operating condition compatibility verification. This mechanism enables the model to quickly call upon historical similar data for validation and calibration when operating conditions change, ensuring the consistency and reliability of prediction results across different operating cycles.

[0061] By dynamically adjusting the mapping relationship between feature weight parameters and the benchmark template through the feedback matrix, the collaborative optimization of all components of the model is achieved. The dynamic updating of feature reconstruction instructions and physical field coupling parameters enables the model to continuously adapt to nonlinear changes during battery degradation, maintaining high prediction accuracy throughout the battery's entire lifespan and providing a more reliable lifespan assessment basis for battery management of electrochemical energy storage systems. Attached Figure Description

[0062] Figure 1 This is a schematic diagram illustrating the working principle of a method for predicting the remaining battery life of an electrochemical energy storage system according to the present invention.

[0063] Figure 2 A flowchart for iterative updating of parameters in a multi-scale model;

[0064] Figure 3 The flowchart is for dynamic weight superposition control;

[0065] Figure 4 A flowchart for constructing the adaptive evaluation function for the model;

[0066] Figure 5 A flowchart for generating the feedback matrix. Detailed Implementation

[0067] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] Please see Figure 1 This invention provides a method for predicting the remaining life of a battery in an electrochemical energy storage system, the method comprising:

[0069] High-precision lifetime prediction is achieved through multi-dimensional data fusion and dynamic modeling. The system first collects multi-dimensional operational data from the target battery pack during charge-discharge cycles, including temperature distribution data, current ripple data, and voltage decay curves. Feature weight parameters are generated through matching analysis between real-time operating conditions and historical databases. A full-lifecycle degradation feature pattern analysis framework is established, generating a benchmark template based on capacity decay trajectories at different charge-discharge rates. A multi-scale coupled lifetime prediction model is constructed, employing an iterative optimization mechanism to drive model parameter updates, synchronously generating feature reconstruction instructions and adjusting physical field coupling parameters. A cross-cycle data verification mechanism verifies the operating condition compatibility of the verification data package. A feedback matrix is ​​constructed by combining the prediction error index generated by the model's adaptive evaluation function with the verification results, dynamically optimizing the mapping relationship between feature weight parameters and the benchmark template.

[0070] Example 1: See Figure 2 The construction and parameter optimization process of a multi-scale coupled lifetime prediction model. The core of this implementation lies in transforming battery degradation characteristics into a computable high-dimensional data structure and continuously optimizing model parameters through an iterative mechanism to ultimately achieve accurate prediction of remaining battery lifetime.

[0071] The system first performs tensor mapping processing on the collected battery degradation characteristics. This process converts multi-dimensional operating parameters such as temperature distribution, current ripple, and voltage decay into mathematical expressions with clear physical meaning. Each dimension corresponds to physical field coupling parameters, which define the boundaries of the battery's electrochemical behavior under different operating conditions. The tensor structure considers the interactions between parameters, enabling the model to simultaneously handle multiple interrelated degradation factors. During charge-discharge cycles, the system calculates the correlation coefficients of electrochemical parameters in real time; these coefficients reflect the relative influence of different parameters on the battery degradation rate.

[0072] The construction of the error neighborhood table is a crucial step in this implementation. This data structure stores historical failure cases similar to the current battery state, including the corresponding combinations of physical field parameters and the resulting failure modes. The system uses a clustering algorithm to classify these historical cases, identifying groups of cases with similar degradation trajectories. Each case group is associated with a set of candidate reconstruction instructions, which contain parameter adjustment suggestions that may improve the model's prediction accuracy. The formation of the candidate instruction set considers the historical effects of parameter adjustments and the compatibility relationships between different parameter combinations.

[0073] A dual-constraint screening mechanism is responsible for selecting the optimal adjustment scheme from numerous candidate instructions. This mechanism operates on two complementary dimensions: firstly, it evaluates the performance of each candidate instruction in historical amendment examples, quantifying its actual effect on reducing prediction errors; secondly, it examines the degree of matching between the candidate instructions and the current electrochemical conversion rules, ensuring that parameter adjustments do not violate fundamental physicochemical laws. The screening process employs a dynamic weight allocation strategy, automatically adjusting the relative importance of the two dimensions based on the model's current prediction state. When the model's prediction results fluctuate significantly, the system prioritizes parameter compatibility constraints; during relatively stable prediction phases, it places greater emphasis on error reduction.

[0074] The execution of the target reconstruction command involves multiple coordinated operations. The system first parses the command content to determine the physical field coupling parameters that need adjustment and their target value range. The parameter adjustment process employs a gradual strategy, progressively approaching the optimal value through multiple small adjustments to avoid model instability caused by a single large adjustment. After each parameter update, the system reassesses the data relevance in the error neighborhood table, discarding historical cases with low matching degree to the new parameter combination. This dynamic update mechanism ensures that the model always makes predictions based on the most relevant historical data.

[0075] The operating condition matching degree calculation module is responsible for verifying the effectiveness of parameter adjustments. This module compares the updated degradation feature patterns with the validation dataset in multiple dimensions, calculating the degree of matching between the two in each feature dimension. The matching degree evaluation considers not only the similarity of static parameter values ​​but also the consistency of parameter change trends. The system establishes a dedicated feedback channel to transmit matching degree change information to the feature weight adjustment module, forming a closed-loop optimization system. This feedback mechanism allows the model to continuously fine-tune its internal parameters based on actual results, gradually improving prediction accuracy.

[0076] The adjustment of physical field coupling parameters follows constraints. The system maintains a parameter safety boundary database, defining the allowable adjustment range and rate of change for each parameter. A boundary check is performed before each reconstruction command to prevent parameters from entering dangerous regions. For critical parameters, the system also establishes mutual constraints; when a parameter approaches its upper limit, the adjustment range of related parameters is automatically limited. This protection mechanism effectively prevents model failure due to parameter runaway.

[0077] The model iteration process pays special attention to degradation characteristics at different time scales. Short-term fluctuations reflect the instantaneous changes in the battery's state, medium-term trends reflect the phased characteristics of the degradation process, and long-term decay characteristics show the overall aging trajectory of the battery. The system employs a multi-resolution analysis method to process data from these three time scales simultaneously, ensuring that the prediction results reflect both the current state and long-term variation patterns. The fusion of features from different time scales is achieved through a weighted algorithm, with the weight coefficients dynamically adjusted according to the prediction time span.

[0078] The error correction strategy employs a tiered response mechanism. For minor prediction errors, the system performs routine parameter fine-tuning; when moderate errors occur, it triggers the reconstruction of the error neighborhood table and the update of the candidate instruction set; and when significant prediction errors are encountered, it initiates a comprehensive model diagnostic and parameter reset process. This tiered response ensures both daily operational efficiency and adequate contingency plans for abnormal situations.

[0079] Example 2: See Figure 3 This paper describes a dynamic weight superposition algorithm with a dual-constraint screening mechanism and its specific application in a battery life prediction model. The core of this implementation lies in establishing a multi-dimensional decision-making framework and using an intelligent weight allocation strategy to accurately screen reconstruction instructions, thereby optimizing the model parameter adjustment process.

[0080] In the initial stage of system operation, a candidate refactoring instruction evaluation system was constructed, which includes two independent yet complementary evaluation dimensions. The historical correction effect dimension quantifies the performance of each candidate instruction in practical applications by analyzing error correction records within a preset time window. The system counts the number of successful corrections for each instruction, records the magnitude of its reduction in prediction error, and correlates these data with the changing trends of instruction priority. These raw indicators are normalized and transformed into comparable initial screening weights. The processing considers the differences in performance under different battery conditions to ensure the fairness of weight allocation. The initial weight calculation module adopts a sliding time window strategy, giving newer correction records a relatively higher consideration ratio, enabling the system to adapt to changes in battery degradation characteristics.

[0081] The electrochemical compatibility dimension extracts parameter constraints from a benchmark template, reflecting the fundamental physicochemical laws governing battery operation. The system analyzes key parameter fields in the electrochemical conversion rules, establishing a network of constraint relationships between parameters. Candidate reconfiguration instructions are mapped into this network, and the degree of matching between their parameter adjustment schemes and the constraint fields is calculated. Compatibility evaluation not only checks whether parameter values ​​are within allowable ranges but also analyzes whether parameter combinations violate inherent interrelationships. The evaluation results are converted into compatibility constraint weights through a nonlinear mapping function. This function is more sensitive to key parameter settings, significantly reducing the weight of adjustments that may jeopardize battery safety.

[0082] The error neighborhood table analysis module monitors data distribution characteristics in real time, particularly the changes in the proportion of low-correlation data. This module uses a density clustering algorithm to identify sparse regions in the data distribution; these regions typically correspond to high-risk states predicted by the model. The system calculates the error risk level based on the degree of clustering of low-correlation data; a higher risk level indicates greater uncertainty in the current prediction. The risk level calculation uses a multi-segment function, dividing the continuous density index into several risk intervals, each corresponding to a different decision-making strategy. The historical moving average is used as a stability benchmark to determine whether the current risk level is within the normal fluctuation range.

[0083] The dynamic superposition factor generator calculates weight adjustment parameters based on error risk level and stability index. This generator employs a threshold-based segmented adjustment mechanism, setting different adjustment sensitivities for different risk ranges. When the system is in a low-risk state, the superposition factor remains relatively stable, primarily relying on the initial screening weights for decision-making. As the risk level increases, the superposition factor gradually enhances the influence of compatibility constraints, and in extreme cases, it can completely switch to a conservative decision-making mode. The stability factor controls the fluctuation amplitude of the superposition factor, preventing decision oscillations caused by short-term data fluctuations. The deviation of the superposition factor from its historical mean is continuously monitored, and the strength of the stability factor is automatically adjusted when the deviation exceeds a preset threshold.

[0084] The weight fusion process employs a dynamic mixing algorithm, combining the initial screening weights with the compatibility constraint weights according to a ratio determined by the superposition factor. The fusion algorithm considers the nonlinear interactions between weights and determines the optimal mixing method through cross-validation. In certain special cases, when two weight indicators severely conflict, the system initiates a conflict resolution protocol, prioritizing parameters crucial to battery safety. The fusion result generates a comprehensive screening probability for each candidate instruction; these probabilities, after standardization, drive the final selection operation.

[0085] The instruction selection operation employs a strategy combining probability sampling and deterministic checks. The system first randomly pre-selects several candidate instructions according to a comprehensive screening probability distribution, and then performs detailed compatibility verification on these instructions. The verification process checks the interaction between the pre-selected instructions and the core parameters of the baseline template, especially complex adjustment schemes involving coupling multiple physical fields. When parameter conflicts are detected, the system records the conflict type and severity and automatically switches to the backup instruction queue. Instructions in the backup queue have been pre-screened and have a high level of basic compatibility assurance. After the selection operation is completed, the selected target reconstruction instruction enters the execution queue, and the historical selection records of related instructions are updated.

[0086] The conflict detection and handling subsystem continuously monitors the parameter status during command execution. This system maintains a parameter interaction knowledge base, storing information on potential side effects from various parameter combinations. When a potential conflict is detected, the system performs pattern matching based on cases in the knowledge base to quickly identify the conflict type and take corresponding mitigation measures. For new conflict patterns, the system initiates a learning process, adding new conflict characteristics and their handling solutions to the knowledge base. Conflict handling employs a progressive strategy, escalating responses gradually from warnings and parameter fine-tuning to command rollback, based on the severity of the conflict.

[0087] The standby instruction queue management module maintains a pre-screened set of candidate instructions. This module periodically extracts valid instructions from historical success cases and performs compatibility enhancement processing. Newly generated candidate instructions undergo observation and verification in the standby queue for a period of time before entering the primary selection pool. Instructions in the queue are sorted by priority based on metrics such as historical success rate, compatibility score, and recent usage frequency. When the primary selection pool cannot provide a suitable instruction due to conflicts or other reasons, the system selects an alternative from the standby queue according to priority.

[0088] A complete instruction lifecycle management mechanism was established during the system implementation process. Newly generated candidate instructions first enter an observation period, during which they can only be used for low-risk prediction tasks. After thorough verification, they are transferred to the regular use phase. When the correction effect of an instruction continues to decline or compatibility issues arise, it is gradually downgraded until it is phased out. This lifecycle management ensures the dynamic updating of the instruction library, continuously adapting to changes in battery degradation characteristics.

[0089] The data tracking and diagnostic subsystem records complete contextual information for all screening decisions, including the model state at the time, input parameters, weight calculation process, and final selection result. This data is used for subsequent decision quality analysis and algorithm improvement. Diagnostic tools can reproduce historical decision-making scenarios, helping to understand the reasons for specific screening results. When systematic biases or decision-making flaws are detected, the system initiates a calibration process to adjust key parameters in the weight calculation.

[0090] The visual interface provides operators with an interpretable demonstration of the decision-making process. It graphically presents the various evaluation metrics, weighting ratios, and decision path of the final selection results for candidate instructions. The interface supports interactive exploration, allowing users to delve into detailed evaluation data and historical performance records for specific instructions. These visualization tools enhance system transparency and facilitate human oversight and intervention.

[0091] Example 3: See Figure 4 This paper focuses on the construction of the adaptive evaluation function and the generation of dynamic correction factors, emphasizing how to achieve accurate assessment of battery degradation status through spatiotemporal correlation analysis and multi-dimensional data fusion. This implementation establishes a complete parameter correlation calculation system, organically integrating battery operating data, physical field coupling parameters, and degradation characteristics to form an adaptive evaluation framework.

[0092] The system first analyzes the distribution characteristics of the feature weight parameters and uses a clustering algorithm based on kernel density estimation to identify high-density regions in the parameter space. These regions correspond to the typical state of the battery under specific operating conditions, and each cluster center is considered a reference benchmark. Real-time operating data is projected onto this parameter space, and the multidimensional distance between the data and each benchmark is calculated to form a spatiotemporal correlation matrix. Each element in the matrix represents the correlation strength between a specific dimension parameter and the battery degradation rate, taking into account the time delay effect and spatial distribution characteristics during calculation. The correlation calculation employs an improved grey relational analysis method, which can handle nonlinear and non-stationary parameter relationships.

[0093] The generation process of the dynamic correction factor is based on the analysis of electrochemical parameter fluctuations. The system monitors the data streams of high decay rate regions and temperature anomaly regions, aligning their change trajectories using a dynamic time warping algorithm. During trajectory alignment, key inflection points and trends are identified, quantifying the parameter differences between the two regions. The difference calculation considers not only instantaneous value differences but also analyzes the degree of matching of change rates and the persistence of trends. The initial correction factor is generated based on the magnitude and direction of the difference, and its calculation process follows the following relationship:

[0094]

[0095] in, This represents the initial correction factor. The size of the time window for analysis, It is the first The temperature gradient change at each sampling point Corresponding capacity decay gradient, This refers to the local fluctuation range. It is a small constant to prevent division by zero. This formula quantifies the dynamic coupling relationship between temperature anomalies and capacity decay, and normalization ensures that the factor value is within a standard range.

[0096] Historical records of the physical field coupling parameters were organized into a time-series database for constructing the attenuation control model. This model analyzes the attenuation characteristics of the correction factor under different operating conditions and establishes a prediction curve of the factor's effectiveness over time. The attenuation control employs a dynamic damping mechanism, automatically adjusting the damping coefficient based on the current battery state and the model's prediction error. When the prediction error is large, the system reduces the damping strength, allowing the correction factor to play a greater role; when the model tends to stabilize, the damping is appropriately increased to prevent oscillations caused by over-correction.

[0097] The evaluation function is constructed using a multi-layered fusion strategy. The base layer integrates feature weight parameters and spatiotemporal correlation to generate preliminary evaluation values. The intermediate layer introduces a dynamic correction factor to adapt the preliminary evaluation values ​​to the operating conditions. The top layer incorporates a penalty term related to deviations in the physical field coupling parameters, constraining the evaluation results within a reasonable range. The penalty term is calculated using a piecewise function, setting different sensitivities for different types of parameter deviations. Minor deviations only incur linear penalties, while severe deviations in key parameters trigger exponentially increasing penalty strength, forcing the evaluation results to reflect the importance of parameter anomalies.

[0098] The analysis of high degradation rate regions focuses particularly on parameter sensitivity characteristics. The system identifies the contribution of each parameter to the degradation rate through micro-perturbation testing and establishes a sensitivity ranking list. Parameters with high sensitivity receive greater weight in the evaluation function, and their changes significantly affect the final evaluation results. Sensitivity analysis is updated regularly to adapt to changes in parameter importance during battery aging. The detection of abnormal temperature regions employs an adaptive threshold method, with the threshold dynamically adjusted based on historical battery operating data to avoid false alarms or missed alarms caused by fixed thresholds.

[0099] The electrochemical response delay is modeled using the transfer function method. The system performs frequency domain analysis on input and output data from different charge / discharge stages to identify the delay characteristics and amplitude-frequency features of the battery response. These features are quantified as delay parameters to correct the alignment process of the dynamic time warping algorithm. Delay modeling pays particular attention to the phase relationship between temperature change and capacity decay, accurately capturing the hysteresis characteristics of the thermal effect on battery performance.

[0100] Post-processing of the evaluation results includes smoothing filtering and confidence labeling. Smoothing filtering employs a physics-constrained algorithm to eliminate noise while preserving key feature points of the evaluation curve. Confidence calculation considers factors such as data integrity, model fit, and parameter reliability, attaching a quality label to each evaluation point. When the confidence level falls below a threshold, the system automatically triggers a data review process, pausing evaluation output if necessary until data reliability is confirmed.

[0101] The real-time monitoring subsystem continuously tracks the operational status of the evaluation function. This subsystem records the trajectory of evaluation value changes, detecting abnormal fluctuations and trend reversals. When an evaluation value continuously deviates from the normal range, a root cause analysis process is initiated to trace the chain of parameter changes leading to the deviation. The monitoring results are used to optimize the internal parameters of the evaluation function, forming a closed-loop system for continuous improvement.

[0102] The visualization interface transforms the complex evaluation process into intuitive graphical displays. 3D surface plots show the correspondence between evaluation values ​​and key parameters, dynamic heatmaps display state changes in different regions, and trend comparison curves juxtapose evaluation results with actual measurements. The interface supports interactive queries, allowing for in-depth viewing of evaluation details and parameter composition at specific time points.

[0103] Example 4: See Figure 5 The process of generating and applying the feedback matrix is ​​illustrated through specific examples, demonstrating how the system handles the coupling relationship between prediction errors and validation results. The core of this implementation lies in establishing a multi-source data fusion mechanism, organically combining model output, measured data, and historical cases to form a feedback system with self-correcting capabilities.

[0104] Taking a typical lithium-ion battery pack life prediction scenario as an example, the following key parameters were collected during system operation:

[0105] Table 1: Record of battery pack operating parameters and prediction errors.

[0106]

[0107] The system first analyzes the coupling relationship between the error index and the compatibility compliance rate. At time t1, although the error index is low (0.006), the compatibility compliance rate already shows a downward trend (92%). Through historical case matching, the system finds that similar situations in past data often foreshadow an impending increase in error. The coupling coefficient calculation module quantifies this early warning signal into a risk indicator, marking it as a region of concern in the initial feedback matrix.

[0108] At time t3, the temperature rose to 31.8℃, accompanied by a 7.2% increase in current ripple, and the error exponent jumped to 0.015. The system searched the error neighborhood table and found three historical cases similar to this state: Case A ultimately led to prediction failure, Case B recovered stability through parameter adjustments, and Case C entered a period of accelerated decay. The feedback matrix generator analyzed the key differences between these three cases and found that the synchronicity between the voltage decay rate and temperature change was the decisive factor. In the current data, the synchronicity between voltage decay and temperature increase is closer to the characteristics of Case A; therefore, the system marked a high-risk warning in the feedback matrix.

[0109] The verification data packet validation process revealed that the measured capacity decay rate under the current charge / discharge mode exceeded the expected range of the baseline template. The compatibility verification module decomposed this deviation into three components: temperature influence factor, current distribution factor, and aging acceleration factor. By comparing the operating parameters in the verification data packet with historical records, the system identified abnormal current distribution as the primary cause. This finding was encoded into a correction instruction for the feedback matrix, requiring adjustment of the weighting coefficients of the current ripple parameter.

[0110] The update intensity of the feedback matrix is ​​controlled using a case-based learning mechanism. The system analyzes the intervention effects in similar scenarios across historical correction records and finds that moderate parameter adjustments are most effective for compatibility degradation caused by abnormal current distribution. Excessive correction leads to overfitting the current anomaly, while insufficient correction fails to prevent error amplification. Based on this understanding, the system sets the update intensity of the current feedback matrix to a moderate level, corresponding to a parameter weight adjustment range of 20%-30%.

[0111] The construction process of the multidimensional feedback model demonstrates the system's intelligent fusion capability. At time t4, although the temperature dropped somewhat, the error index continued to climb to 0.017. Instead of simply relying on temperature parameters for judgment, the system comprehensively analyzed the changes in the second derivative of the voltage decay curve and the spectral characteristics of the current ripple. These analyses revealed a new accelerating trend in voltage decay, while high-frequency components appeared in the current ripple. The feedback model matched these two characteristics with a historical case library, identifying typical modes of battery separator aging, and thus adding material-level degradation markers to the feedback matrix.

[0112] The strategy for handling high-error regions demonstrates the system's hierarchical decision-making capability. When the error index exceeds the 0.015 threshold, the system automatically activates a three-level response mechanism: the primary response adjusts the model parameter weights, the intermediate response corrects the feature extraction algorithm, and the advanced response triggers a baseline template update. In the current example, the system first attempts to improve the prediction by reallocating the weights of temperature and current parameters, while simultaneously monitoring the error trend of subsequent data points. This gradual response avoids overreaction and effectively controls error expansion.

[0113] The visualization of the feedback matrix employs multi-dimensional projection technology. The user interface displays a three-dimensional coordinate space, with the three axes representing prediction error, compatibility achievement rate, and parameter sensitivity, respectively. Each data point forms a dynamic cloud in the space, with color intensity indicating risk level and shape changes reflecting trend characteristics. The data from time t1 to t4 forms a clear evolution path in the space, helping operators understand the system state transition process.

[0114] The conflict resolution mechanism plays a crucial role in actual operation. When a correction suggestion in the feedback matrix conflicts with the core constraints of the baseline template, the system initiates a negotiated decision-making process. For example, a correction suggestion might be to increase the temperature weight to reduce error, but the safety constraints of the baseline template prohibit increasing the weight's influence in the high-temperature range. The system uses a compromise algorithm to find a feasible solution, appropriately adjusting the temperature weight within permissible limits while simultaneously enhancing the compensation effect of the cooling system parameters. This optimization under constraints demonstrates the system's engineering practicality.

[0115] The system maintenance module ensures the long-term reliability of the feedback mechanism. Regularly executed diagnostic processes check the health status of each element in the feedback matrix, including the timeliness of historical data, the applicability of association rules, and the rationality of weighting coefficients. When the diagnostics detect that an association rule for a certain parameter has become invalid due to improvements in battery technology, the system automatically initiates a rule update process, learning from the latest data to build a new association model. This self-maintenance function ensures that the system can adapt to the iterative development of battery technology.

[0116] The anomaly handling process is designed to consider various boundary conditions in actual operation. When a sensor malfunction leads to data loss, the system can generate reasonable replacement values ​​and mark data quality based on the characteristics of adjacent data points and the overall state of the battery pack. For sudden abnormal events such as current surges, the system will temporarily freeze the update of the feedback matrix, waiting to confirm the nature of the event before deciding on a handling strategy. These design details ensure the stable operation of the system under non-ideal conditions.

[0117] Integration with external systems is achieved through standardized interfaces. The output format of the feedback matrix is ​​compatible with common battery management system protocols, supporting collaborative operation with subsystems such as charge / discharge control devices and thermal management units. In the example scenario, when the system detects a high-temperature condition accompanied by high error, it not only adjusts the parameters of the internal prediction model but also sends optimization commands to the cooling system via a digital interface, forming a cross-system joint optimization scheme.

[0118] Example 5: This example focuses on the quantitative model of battery pack degradation state and topology network analysis, detailing how to achieve accurate matching between battery operating state and historical degradation patterns using a dynamic time warping algorithm. This implementation establishes a complete feature space analysis method, transforming the degradation process throughout the battery's lifecycle into a computable topology, providing deep state recognition capabilities for the prediction model.

[0119] The system initialization phase constructs the basic architecture of the degradation feature space, which organizes historical degradation data in a hierarchical network. Each node in the network represents a degradation state, containing a complete set of feature vectors for that state, covering multi-dimensional parameters such as temperature distribution, current characteristics, and voltage response. The connections between nodes encode state transition probabilities, reflecting the typical path of the battery transitioning from one degradation stage to another. An adaptive clustering algorithm is used during network construction to automatically determine the number and location of nodes, ensuring coverage of degradation patterns under various typical operating conditions.

[0120] The data preprocessing module segments the real-time acquired runtime data using a sliding window, with the window size dynamically adjusted based on the battery's charge-discharge cycles. Each data window is assigned a unique timestamp index, recording its position within the battery's lifespan. The window overlap rate is set as an adjustable parameter, striking a balance between data processing accuracy and computational efficiency. The segmented data fragments undergo standardization to eliminate differences caused by varying sensor dimensions and sampling frequencies, establishing a unified benchmark for subsequent feature comparison.

[0121] The electrochemical feature extraction process employs a multi-resolution analysis method to characterize the essential properties of data fragments from both the time and frequency domains. Time-domain analysis focuses on parameter variation trends, identifying key inflection points and stable intervals; frequency-domain analysis reveals periodic fluctuations and anomalous harmonic components. Feature vector construction considers the phase relationships between different parameters, capturing patterns of coordinated changes in multiple physical quantities. After dimensionality reduction, the extracted feature vectors form standardized state descriptors, facilitating similarity calculations with topological network nodes.

[0122] The dynamic time warping algorithm is specifically optimized for the characteristics of battery data. The core of the algorithm lies in establishing the optimal matching path between the current data segment and historical node features, allowing for flexible alignment of similar shapes along the timeline. Path search employs a dynamic programming method under constraints, limiting the maximum time scaling range and preventing forced matching that does not conform to physical meaning. Similarity calculation introduces a composite metric of shape distance and trend distance, considering both the proximity of instantaneous values ​​and the consistency of the direction of change.

[0123] The similarity matrix is ​​constructed using an incremental update strategy, gradually improving as new data arrives. Each element in the matrix records the degree of matching between a specific data segment and a topological node, with color coding visually displaying the distribution of matching quality. When the backtracking algorithm extracts the optimal path from the matrix, it comprehensively considers both local matching degree and global continuity, avoiding the selection of paths that, while locally well-matched, are inconsistent overall. Path smoothing eliminates minor fluctuations caused by measurement noise, highlighting the true degradation trend.

[0124] The generation of model correction commands is based on the quantitative analysis of path deviation. The system continuously monitors the deviation between the current optimal path and the reference path, which is derived from typical trajectories of the same degradation stage in historical data. Deviation calculation distinguishes between systematic offsets and random fluctuations; the former may indicate substantial changes in battery performance, while the latter usually reflects temporary disturbances. When a systematic offset exceeds a preset threshold, different levels of correction commands are triggered, ranging from parameter fine-tuning to model structure updates, forming a graded response mechanism.

[0125] Dynamic maintenance of the network topology is a crucial aspect of the implementation process. The system periodically assesses the impact of new data on the network structure, and initiates a network expansion process when a new degradation pattern is found that cannot be fully represented by existing nodes. The insertion position of new nodes is determined by the density distribution of existing nodes, prioritizing the filling of sparse regions. A node merging mechanism handles similar nodes that become redundant with data accumulation, maintaining network compactness and expressive efficiency. The weights of connection edges are dynamically updated based on the latest data, reflecting changes in the probability of degradation paths during battery aging.

[0126] The visualization of state recognition results employs multidimensional projection technology, mapping key dimensions of the high-dimensional feature space to a two-dimensional or three-dimensional display plane. The projection algorithm preserves the topological relationships in the original space, ensuring that similar states still cluster after projection. The positional changes of the current data segment in the projected image form a dynamic trajectory, allowing operators to intuitively observe the migration process of battery states in the feature space. The trajectory playback function supports retrospective analysis, helping to understand the causes and development trends of specific states.

[0127] The anomaly handling mechanism covers the entire process of data acquisition, feature extraction, and state matching. The sensor fault detection module identifies abnormal readings and attempts to compensate appropriately using adjacent data and model predictions. Outliers during feature extraction are handled using robust statistical methods to reduce their interference with the overall analysis. In the state matching phase, low-quality data segments are specially marked to avoid incorrect state determinations due to data issues. The system maintains an anomaly case knowledge base for quickly identifying and responding to recurring problem patterns.

[0128] Computational efficiency optimizations are implemented at multiple levels. Feature extraction employs incremental computation, processing only the changed portions of data as new data arrives. Topology network search utilizes hierarchical indexing technology to quickly locate subsets of potentially matching nodes. Dynamic time warping implements various acceleration strategies, including early termination, approximate search, and parallel computation. Memory management employs an intelligent caching mechanism, optimizing storage layout based on data access patterns. These optimizations enable the system to run in real-time on resource-constrained hardware platforms.

[0129] The interface design with the predictive model supports bidirectional data flow. State identification results are passed as important input features to the predictive model, enhancing its understanding of the current battery degradation stage. Simultaneously, the predictive model's output is fed back into the state identification process, helping to verify and correct the feature space boundaries. This tight integration enables the two modules to evolve collaboratively, jointly adapting to changes in battery performance with aging.

[0130] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof 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 process, method, article, or apparatus.

[0131] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for predicting the remaining lifespan of a battery in an electrochemical energy storage system, characterized in that, include: Multi-dimensional operating data of the target battery pack during the charge and discharge cycle are collected. The multi-dimensional operating data includes temperature distribution data, current ripple data and voltage decay curve. Feature weight parameters are generated based on the matching results of real-time operating conditions and historical operating database. Degradation feature patterns are extracted throughout the entire life cycle of the target battery pack. These degradation feature patterns are generated by analyzing the capacity decay trajectory under different charge and discharge rates, and are used as a benchmark template for establishing a dynamic degradation feature space. A multi-scale coupled lifetime prediction model is constructed, driving the model parameters to iteratively update in the direction of minimizing prediction error. Each iteration generates feature reconstruction instructions and updates the physical field coupling parameters of the benchmark template. An adaptive evaluation function for the model is constructed based on the correlation between feature weight parameters and real-time running data. Cross-cycle data verification is triggered based on the amplitude of the feature reconstruction instruction. The cross-cycle data verification extracts a verification data packet that matches the current charge and discharge mode from the historical operation database and performs operating condition compatibility verification on the verification data packet through the electrochemical conversion rules in the benchmark template. The prediction error index output by the model adaptive evaluation function and the verification data packet verification results are used to generate a feedback matrix, and the mapping relationship between the feature weight parameters and the benchmark template is dynamically adjusted. The adaptive evaluation function for the model, constructed based on the correlation between feature weight parameters and real-time running data, includes: Based on the distribution characteristics of the feature weight parameters, the spatiotemporal correlation between the parameters of each dimension and the battery degradation rate in the real-time running data is calculated. A dynamic correction factor is generated based on spatiotemporal correlation, which correlates the parameter change trends of high decay rate regions with temperature anomaly regions, and constrains the scope of the dynamic correction factor through the physical field coupling parameters of the benchmark template. By integrating feature weight parameters, dynamic correction factors, and spatiotemporal correlation, an adaptive evaluation function for the model is constructed, and a penalty term related to the deviation of the physical field coupling parameters is introduced.

2. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 1, characterized in that, The construction of a multi-scale coupled lifetime prediction model, driving the model parameters to iteratively update in the direction of minimizing prediction error, includes: The degradation feature patterns are mapped to high-dimensional tensors, with each dimension corresponding to the adjustable range of physical field coupling parameters. The correlation coefficients of electrochemical parameters are calculated based on the current charge and discharge modes. Based on the correlation coefficient, an error neighborhood table is constructed to record historical failure cases that are associated with the current degradation feature mode and their differences in physical field coupling parameters, thereby generating a candidate reconstruction instruction set. A dual-constraint screening mechanism is used to select target reconstruction instructions from the candidate reconstruction instruction set. The constraints include the error reduction rate of the candidate reconstruction instructions in historical corrections and the parameter compatibility threshold between the candidate reconstruction instructions and the electrochemical conversion rules. The target reconstruction instruction is applied to the degenerate feature pattern to dynamically update the parameter values ​​of the corresponding dimension in the high-dimensional tensor, while adjusting the filtering threshold of the error neighborhood table to trigger the elimination of low-relevance data. The process involves recalculating the matching degree between the updated degraded feature pattern and the verification data packet, and feeding back the change in matching degree to the adjustment of the feature weight parameters.

3. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 2, characterized in that, The step of selecting target reconstruction instructions from the candidate reconstruction instruction set using a dual-constraint screening mechanism includes: The error reduction rate of each candidate reconstruction instruction within the preset time window is statistically analyzed, and the ratio of the number of historical successful corrections to the change in the priority of reconstruction instructions is normalized to the initial screening weight. Extract the parameter constraint fields corresponding to the electrochemical conversion rules in the benchmark template, calculate the coverage ratio of the parameter dimensions corresponding to the candidate reconstruction instructions and the preset constraint fields, and map them as compatibility constraint weights; The initial screening weights and compatibility constraint weights are dynamically superimposed, and the superposition coefficient is adjusted according to the distribution density of low-correlation data in the error neighborhood table, so that the compatibility constraint weights can obtain a higher proportion in the high-risk area of ​​prediction error. Based on the comprehensive screening probability, the instruction selection operation is performed to verify the compatibility between the candidate reconstruction instructions and the core parameters of the baseline template. If there is a conflict, the backup reconstruction instruction queue is activated.

4. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 3, characterized in that, The step of dynamically superimposing the initial screening weights and compatibility constraint weights includes: Based on the comparison between the proportion of low-relevance data in the error neighborhood table and the historical maximum data volume, the error risk level is calculated through a multi-segment function. A dynamic superposition factor is generated based on the error risk level. A threshold segmented adjustment mechanism is adopted, and the historical moving average is used as a stability factor to control the fluctuation range of the dynamic superposition factor. The initial screening weights are correlated with the dynamic superposition factors, and the correlation results are fused with the compatibility constraint weights. The strength of the stability factor is adjusted based on the degree of deviation between the dynamic superposition factor and the historical moving average.

5. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 1, characterized in that, The generation of dynamic correction factors based on spatiotemporal correlation includes: Based on the spatiotemporal correlation and the inverse correlation between electrochemical parameter fluctuations and capacity decay gradient, the difference in parameter changes between the high decay rate region and the temperature anomaly region is calculated. An initial correction factor is generated based on the difference in parameter changes. The parameter sensitivity in the high decay rate region is correlated with the electrochemical response delay in the temperature anomaly region. The rate of change of the correction factor is constrained by the physical field coupling parameters of the reference template. Based on the historical correction records of the physical field coupling parameters, an attenuation control factor is applied to the initial correction factor.

6. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 1, characterized in that, The step of generating a feedback matrix by combining the prediction error index output by the model adaptive evaluation function with the verification data packet validation results includes: The error-compatibility coupling coefficient is calculated based on the prediction error index and the compatibility compliance rate in the verification data packet results. An initial feedback matrix is ​​generated based on the error and compatibility coupling coefficient. The lack of compatibility in high error regions is associated with feature reconstruction instructions. The update strength of the constraint feedback matrix is ​​constrained by the historical correction record of the benchmark template. A multidimensional feedback model is constructed by fusing the coupling coefficient between error and compatibility, the prediction error index, and the compatibility achievement rate.

7. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 1, characterized in that, Also includes: Establish a quantitative model of battery pack degradation state and analyze the main and secondary feature quantities in multi-dimensional operating data; Construct a degradation feature space topology network and encode the historical degradation patterns of different charge and discharge stages as topology nodes; The similarity path between the current running data and the topological nodes is calculated using a dynamic time warping algorithm. When the similarity path exceeds the preset threshold, the prediction model parameter reset mechanism is triggered.

8. The method for predicting the remaining battery life of an electrochemical energy storage system according to claim 7, characterized in that, The step of calculating the similarity path between the current running data and the topological nodes using the dynamic time warping algorithm includes: The current running data is segmented using a sliding window to generate multiple data fragments and their timestamp indices. Extract the electrochemical feature vector of each data segment, and calculate the dynamic bending path between the electrochemical feature vector of the data segment and the feature vector of the topological node; A similarity matrix is ​​generated based on a dynamic curved path, and the optimal matching path is located through a backtracking algorithm. Model correction instructions are generated based on the degree of deviation between the optimal matching path and the preset threshold.

9. A system for predicting the remaining life of an electrochemical energy storage system battery, used to implement the method for predicting the remaining life of an electrochemical energy storage system battery as described in any one of claims 1-8, characterized in that, The system includes: The data acquisition module collects multi-dimensional operating data of the target battery pack during the charge and discharge cycle, and generates feature weight parameters based on the matching results of real-time operating conditions and historical operating database. The template construction module extracts the degradation feature patterns throughout the entire life cycle of the target battery pack and uses these degradation feature patterns as a reference template for establishing a dynamic degradation feature space. The model iteration module constructs a multi-scale coupled lifetime prediction model, drives the model parameters to be updated iteratively in the direction of minimizing the prediction error, and constructs an adaptive evaluation function for the model based on the correlation between feature weight parameters and real-time running data. The verification trigger module triggers cross-cycle data verification based on the amplitude of the feature reconstruction instruction, and performs operating condition compatibility verification on the verification data package through the electrochemical conversion rules in the benchmark template. The feedback generation module generates a feedback matrix by combining the prediction error index output by the model's adaptive evaluation function with the verification results of the validation data package, and dynamically adjusts the mapping relationship between the feature weight parameters and the benchmark template.

Images