Low-temperature performance repairing method of lithium iron phosphate battery

Abstract

The invention discloses a low-temperature performance repairing method for a lithium iron phosphate battery, and relates to the technical field of battery performance repairing, and the method comprises the steps: obtaining low-temperature application scene information of a target lithium iron phosphate battery, and carrying out the dynamic optimization of the battery, and obtaining a battery structure optimization parameter system; carrying out low-temperature twinborn simulation to generate a battery low-temperature operation twinborn body; constructing a battery low-temperature performance evaluation index, and generating a battery low-temperature operation prediction performance parameter; and carrying out repair strategy analysis, constructing a battery low-temperature performance repair strategy scheme library, and carrying out battery operation control. The technical problems that in the prior art, lithium iron phosphate battery low-temperature performance optimization lacks scene adaptability, performance evaluation and repair control are disjointed, and consequently the low-temperature operation stability is poor and the performance recovery efficiency is low are solved, and the purposes that whole-process accurate intervention of the lithium iron phosphate battery low-temperature performance is achieved, and the low-temperature performance of the lithium iron phosphate battery is improved are achieved. And the operation stability and the performance recovery efficiency of the battery in a low-temperature scene are improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of battery performance repair, in particular to a low-temperature performance repair method for a lithium iron phosphate battery. BACKGROUND

[0002] The lithium iron phosphate battery is widely applied to the fields of new energy vehicles and energy storage systems due to the advantages of low cost, high safety and long cycle life, but the performance and reliability of the lithium iron phosphate battery are significantly restricted in a low-temperature environment. The ion conductivity of electrolyte and the electrode reaction kinetics are reduced in the low-temperature environment, which easily leads to a significant capacity attenuation of the battery, a reduction of the charging and discharging efficiency, and even causes safety hazards such as lithium precipitation and internal resistance increase, and seriously affects the normal operation of equipment. The current improvement methods for the low-temperature performance of the lithium iron phosphate battery are mostly concentrated in single-dimension optimization, such as improving the basic low-temperature characteristics through static technology such as electrolyte formula adjustment and electrode material modification, or relying on a simple heating and insulation device for passive temperature control. Such methods generally lack the adaptability to the actual low-temperature application scenarios of the battery and are difficult to realize dynamic evaluation and accurate repair of the performance in combination with the historical operation data of the battery, which leads to limited performance optimization effect and insufficient targetedness of the repair strategy.

[0003] The prior art has the technical problems of lacking scenario adaptability in the low-temperature performance optimization of the lithium iron phosphate battery, and the performance evaluation and repair control are disconnected, leading to poor running stability and low performance recovery efficiency in the low-temperature environment. SUMMARY

[0004] The application provides a low-temperature performance repair method for a lithium iron phosphate battery, which is used to solve the technical problems of lacking scenario adaptability in the low-temperature performance optimization of the lithium iron phosphate battery in the prior art, and the performance evaluation and repair control are disconnected, leading to poor running stability and low performance recovery efficiency in the low-temperature environment.

[0005] In view of the above problems, the application provides a low-temperature performance repair method for a lithium iron phosphate battery.

[0006] The application provides a low-temperature performance repair method for a lithium iron phosphate battery, which comprises the following steps: The low-temperature application scenario information of a target lithium iron phosphate battery is acquired, the battery is dynamically optimized based on the low-temperature application scenario information, and a battery structure optimization parameter system is obtained; a battery historical low-temperature operation data set is collected through a BMS, low-temperature twin simulation is performed based on the battery structure optimization parameter system and the battery historical low-temperature operation data set, and a battery low-temperature operation twin is generated; a battery low-temperature performance evaluation index is constructed, battery performance prediction is performed on the battery low-temperature operation twin according to the battery low-temperature performance evaluation index, and a battery low-temperature operation predicted performance parameter is generated; repair strategy analysis is performed on the battery low-temperature operation predicted performance parameter, a battery low-temperature performance repair strategy scheme library is constructed, and battery operation control is performed based on the battery low-temperature performance repair strategy scheme library.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: This method involves acquiring low-temperature application scenario information for the target lithium iron phosphate battery, performing dynamic battery optimization, and obtaining a battery structure optimization parameter system. Historical low-temperature operation datasets of the battery are collected through a Battery Management System (BMS), and low-temperature twin simulations are performed to generate a low-temperature operation twin of the battery. Low-temperature performance evaluation indicators for the battery are constructed, and predicted performance parameters for low-temperature operation are generated. Repair strategies are analyzed based on these predicted performance parameters, and a library of low-temperature performance repair strategies is constructed for battery operation control. This approach achieves precise intervention throughout the entire low-temperature performance process of lithium iron phosphate batteries, improving the battery's operational stability and performance recovery efficiency in low-temperature scenarios. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A schematic flowchart of a method for repairing the low-temperature performance of a lithium iron phosphate battery provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the process of obtaining the battery structure optimization parameter system in the low-temperature performance repair method for lithium iron phosphate batteries provided in this application embodiment. Detailed Implementation

[0010] This application provides a method for repairing the low-temperature performance of lithium iron phosphate batteries, which addresses the technical problems in the prior art where the low-temperature performance optimization of lithium iron phosphate batteries lacks scenario adaptability, and the performance evaluation and repair control are disconnected, resulting in poor operational stability and low performance recovery efficiency at low temperatures.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] Examples, such as Figure 1 As shown, this application provides a method for repairing the low-temperature performance of lithium iron phosphate batteries, the method comprising: Step S100: Obtain low-temperature application scenario information of the target lithium iron phosphate battery, and perform dynamic optimization of the battery based on the low-temperature application scenario information to obtain a battery structure optimization parameter system.

[0013] Specifically, through scenario research and data collection, key information on the low-temperature application scenarios of the target battery is obtained, including the lowest temperature of the actual working environment, such as -20℃ and -30℃, the discharge rate requirements under typical low-temperature conditions, such as 1C and 0.5C, continuous operation time, low-temperature storage period, and whether frequent start-stop is required. Then, a pre-set lithium iron phosphate battery formula database is accessed, containing data on positive electrode material ratios, electrolyte composition, and negative electrode modification schemes, as well as a battery structure database, containing data on the number and layout of tabs, heat dissipation channel design, and cell stacking methods. The collected scenario information is matched and optimized with the parameters in the two databases in multiple dimensions. For example, for the -30℃ low-temperature and 1C discharge scenario, high-conductivity electrolyte formulas and multi-tab structure designs are prioritized to obtain optimized battery formula data and optimized battery structure data suitable for this scenario. Finally, based on these optimized data, the original formula and structural parameters of the target lithium iron phosphate battery, such as adjusting the lithium salt concentration in the electrolyte and optimizing the tab spacing, are dynamically adjusted and compliance verified, integrating them to form a battery structure optimization parameter system covering formula optimization parameters and structural optimization parameters.

[0014] Step S200: Collect historical low-temperature operation data of the battery through BMS, perform low-temperature twin simulation based on the battery structure optimization parameter system and the historical low-temperature operation data of the battery, and generate a low-temperature operation twin of the battery.

[0015] Specifically, the battery management system (BMS) extracts comprehensive operational data of the target lithium iron phosphate battery under low-temperature conditions, covering charge and discharge voltage curves, real-time current changes, cell temperature distribution, capacity decay rate, and voltage platform stability at different low-temperature ranges such as -10℃, -20℃, and -30℃. This data is then integrated to form a historical low-temperature operation dataset. Subsequently, outlier cleaning is performed on this dataset to remove jump data caused by sensor malfunctions, and standardization is implemented to unify data format and units. Simultaneously, based on the obtained battery structure optimization parameter system, including formulation optimization parameters such as electrolyte composition and structural optimization parameters such as heat dissipation channel dimensions, the dataset is optimized. The process involves using 3D modeling tools to construct a basic battery model that perfectly matches the structure and material properties of the actual battery. Next, a task list is built that includes electrochemical prediction tasks, such as lithium-ion insertion / extraction rate and internal resistance change prediction, and thermal conduction prediction tasks, such as temperature conduction efficiency and local hotspot distribution. Based on the processed historical low-temperature operating dataset, a corresponding battery low-temperature prediction task model set is trained and generated. This model set is then coupled with the basic battery model to simulate the electrochemical reactions and thermal conduction processes of the battery under different low-temperature conditions. Low-temperature twin simulations are performed, ultimately generating a battery low-temperature operating twin that can map the battery's low-temperature operating state in real time and accurately reflect the performance change patterns.

[0016] Step S300: Construct battery low-temperature performance evaluation index, predict battery performance of the battery low-temperature operation twin according to the battery low-temperature performance evaluation index, and generate battery low-temperature operation prediction performance parameters.

[0017] Specifically, focusing on the core requirements of lithium iron phosphate batteries for low-temperature operation, a battery low-temperature performance evaluation index is constructed, encompassing three dimensions: electrochemical performance, including low-temperature capacity retention, discharge efficiency, and voltage plateau voltage drop; thermal management performance, including cell temperature distribution uniformity, temperature rise rate after low-temperature start-up, and heat dissipation channel thermal conductivity; and safety performance, involving the risk level of lithium plating during low-temperature charging, internal resistance growth rate, and cycle stability decay rate. This index system is then hierarchically broken down, refining each dimension into directly calculable sub-level indicators. For example, electrochemical performance is broken down into -20℃ capacity retention, -30℃ capacity retention, and -40℃ capacity retention. The system considers factors such as 1C discharge efficiency at ℃, and prioritizes low-temperature application scenarios (e.g., capacity retention for outdoor energy storage and start-up heating rate for vehicle applications) to assign weight factors to each sub-level indicator, forming a set of battery-level sub-level indicators and a set of weight factors for these indicators. Finally, based on this indicator set and weight configuration, the system performs full-condition performance simulation and prediction on the battery's low-temperature operating twin. By replicating the electrochemical reactions and heat conduction processes in the twin, the system calculates the specific values ​​of each sub-level indicator and integrates them with the weight factors to obtain a quantitative result reflecting the battery's overall low-temperature performance. This generates predicted performance parameters for battery low-temperature operation that include performance data from various dimensions.

[0018] Step S400: Analyze the repair strategy for the predicted low-temperature operation performance parameters of the battery, construct a battery low-temperature performance repair strategy scheme library, and perform battery operation control based on the battery low-temperature performance repair strategy scheme library.

[0019] Specifically, for predicted performance parameters of batteries operating at low temperatures, such as insufficient low-temperature capacity retention, slow heating rate, and excessive lithium plating risk, repair strategy analysis is conducted. Based on the type of parameter anomaly, corresponding repair strategy parameter thresholds are determined, such as electrolyte temperature control range, charging current limit range, and insulation layer power adjustment threshold. Within these threshold ranges, multiple sets of battery repair strategy parameters are randomly selected, and each set is input into a battery low-temperature operating twin for repair effect simulation evaluation. Quantitative results such as capacity recovery rate, thermal management efficiency, and safety risk reduction corresponding to different parameters are obtained. Based on these results, the repair strategy parameters are globally optimized, first selecting... The N parent parameters with better repair effects are expanded into a parameter population through crossover mutation. The population is then updated through iterative learning until it meets the convergence requirements, forming N battery repair strategy parameter clusters. The comprehensive repair effect of each cluster is then evaluated through twins, and the parameter cluster with the best effect is integrated into a battery low-temperature performance repair strategy scheme library. Finally, the target strategy with the best repair effect is determined based on the scheme library and applied to the actual operation control of the battery. If the actual repair effect of the strategy does not reach the preset low-temperature performance standard, such as the capacity retention rate not reaching 80%, the second-best strategy in the scheme library is immediately activated to replace it, ensuring that the battery always maintains a stable operating state in the low-temperature environment.

[0020] In one possible implementation, such as Figure 2 As shown, step S100 further includes: Step S110: Construct a lithium iron phosphate battery formula database and a lithium iron phosphate battery structure database.

[0021] Step S120: Match and optimize the lithium iron phosphate battery formula database and lithium iron phosphate battery structure database with the low-temperature application scenario information to obtain optimized battery formula data and optimized battery structure data.

[0022] Step S130: Based on the battery formulation optimization data and battery structure optimization data, dynamically optimize the formulation and structure data of the target lithium iron phosphate battery to obtain battery formulation optimization parameters and battery structure optimization parameters.

[0023] Step S140: Obtain the battery structure optimization parameter system based on the battery formula optimization parameters and battery structure optimization parameters.

[0024] Specifically, a lithium iron phosphate battery formulation database and a structure database covering multiple types of core information are constructed. The formulation database stores the ratio of cathode materials with different low-temperature adaptability, such as the ratio of LiFePO4 to carbon nanotube conductive agent; electrolyte composition, such as the combination scheme of lithium salts of different concentrations and low-temperature solvents; and anode modification process, such as graphite surface coating parameters. The structure database includes cell structure designs adapted to low-temperature scenarios, such as the number and spacing of tabs, stacking / winding process parameters, heat dissipation system layout, such as heat dissipation channel diameter, insulation layer material parameters, and shell protection structure, such as the thickness of the shell to prevent low-temperature cracking. This provides data support for subsequent matching and optimization.

[0025] Next, the acquired low-temperature application scenario information, such as continuous discharge at -25℃ and 1C discharge rate requirements, is matched with two types of databases in multiple dimensions: In the formulation database, electrolyte formulations that can improve low-temperature ion conduction efficiency are prioritized, such as electrolytes containing ester mixed solvents and positive electrode ratios that enhance electronic conductivity; In the structure database, multi-tab structures that can optimize low-temperature heat distribution and high-density insulation layer designs that reduce heat loss are matched, thereby obtaining battery formulation optimization data and battery structure optimization data that are highly adapted to the scenario requirements.

[0026] Subsequently, based on the original formulation structure data of the target lithium iron phosphate battery, and combined with the formulation optimization data, the core parameters were adjusted, such as optimizing the electrolyte lithium salt concentration from 1 mol / L to 1.2 mol / L and adjusting the positive electrode conductive agent ratio to 2%. The physical structure parameters were optimized according to the structure optimization data, such as increasing the number of tabs from 2 to 4 and adjusting the heat dissipation channel spacing to 5 mm. At the same time, through parameter compatibility verification, it was ensured that the optimized electrolyte and electrode materials did not have adverse reactions. The dynamic optimization configuration was completed, and the final battery formulation optimization parameters and battery structure optimization parameters were obtained.

[0027] Finally, the battery formulation optimization parameters, including parameters related to the positive electrode, electrolyte, and negative electrode, and the battery structure optimization parameters, including parameters related to the cell, heat dissipation, and casing, are classified and integrated. The value range, priority, and applicable scenarios of each parameter are clarified, forming a clear battery structure optimization parameter system that can be directly used for subsequent twin simulation and performance optimization.

[0028] In one possible implementation, step S200 further includes: Step S210: Construct a battery low-temperature prediction task list, which includes battery electrochemical prediction task types and battery thermal conduction prediction task types.

[0029] Step S220: Perform classification and prediction training on the battery historical low temperature operation dataset according to the battery low temperature prediction task list to generate a battery low temperature prediction task model set.

[0030] Step S230: Perform three-dimensional modeling based on the battery structure optimization parameter system to construct the battery basic model.

[0031] Step S240: Couple the battery low-temperature prediction task model set to the battery basic model to perform low-temperature twin simulation and generate a battery low-temperature operation twin.

[0032] Specifically, considering the performance limitations and operational characteristics of lithium iron phosphate batteries in low-temperature environments, a list of battery low-temperature prediction tasks is constructed, comprising two core types. The battery electrochemical prediction task type focuses on the dynamic changes of internal electrochemical reactions within the battery, covering sub-tasks such as predicting the lithium-ion insertion / extraction rate between positive and negative electrode materials under low-temperature conditions, predicting the trend of battery internal resistance changes with temperature and cycle count, predicting capacity decay patterns at different discharge rates, and predicting voltage plateau stability. This is used to accurately capture the impact of low temperatures on battery electrical performance. The battery thermal conduction prediction task type emphasizes the battery's heat transfer and temperature distribution characteristics, including sub-tasks such as predicting the real-time temperature field distribution in various areas of the cell under low-temperature conditions, predicting the thermal conductivity and heat loss of heat dissipation channels, predicting the generation location and temperature peak of local hot spots during charging and discharging, and predicting the heat retention effect of insulation structures. This aims to comprehensively understand the battery's thermal management status under low temperatures. Together, these two types of tasks constitute a prediction system covering the electrochemical-thermal conduction coupling process, providing clear objectives for subsequent model training and twin simulations.

[0033] The historical low-temperature operation dataset of batteries collected by BMS is preprocessed. An outlier detection algorithm, i.e., the 3σ principle, is used to remove invalid data such as voltage jumps and temperature anomalies caused by sensor malfunctions. Then, Min-Max standardization is applied to uniformly map parameters of different magnitudes, such as current, voltage, and temperature, to the [0, 1] interval, resulting in a standardized usable dataset. Subsequently, according to the classification logic of electrochemical prediction and thermal conduction prediction in the battery low-temperature prediction task list, the usable dataset is decomposed into two types of task samples: an electrochemical prediction sample set containing charge / discharge voltage curves, capacity changes, internal resistance data, etc., at different low temperatures, corresponding to sub-tasks such as lithium-ion deintercalation and capacity decay, and a thermal conduction prediction sample set. The test sample set includes data on temperature distribution, heat dissipation power, and the difference between ambient temperature and battery temperature in various regions of the battery cell, corresponding to sub-tasks such as temperature field distribution and heat loss. Then, an appropriate model structure is selected for the characteristics of different sub-tasks. The electrochemical sub-task uses a Long Short-Term Memory (LSTM) network to capture the temporal variation pattern, while the thermal conduction sub-task uses a Convolutional Neural Network (CNN) combined with a finite element analysis model to improve the spatial distribution prediction accuracy. The decomposed sample set is used as training data, and the model training and validation are completed by adjusting hyperparameters such as learning rate and number of iterations. Finally, all the trained sub-task models are integrated to generate a battery low-temperature prediction task model set that can cover the full prediction needs of electrochemistry and thermal conduction.

[0034] Based on a defined battery structure optimization parameter system, key parameters directly related to the battery's physical structure are first extracted. These include the cell's length, width, and height; the thickness and material distribution ratio of the positive / negative electrode / separator; the number, location, and size of the tabs; the diameter, spacing, and layout of the heat dissipation channels; the material parameters and thickness of the insulation layer; and the protective structural dimensions of the battery casing. Then, professional 3D modeling tools, such as ANSYS and SolidWorks, are used to create a full-structure replica of the battery at a 1:1 scale. First, macroscopic structural models of the battery casing and individual cells are constructed. Then, the internal electrode stacking method, electrolyte filling area, and spatial distribution of heat dissipation channels and insulation layers are refined layer by layer. This ensures that the model not only matches the physical battery in external morphology but also accurately reproduces the internal material properties and structural layout, such as the connection method between the tabs and electrodes, and the contact area between the heat dissipation channels and the cells. During the modeling process, material property data from the structural optimization parameter system are simultaneously embedded, such as the conductivity of the electrode materials and the thermal conductivity of the electrolyte. Finally, a basic battery model with both structural realism and parameter accuracy is formed, supporting subsequent twin simulations.

[0035] A bidirectional data interaction interface is established between the battery low-temperature prediction task model set and the battery basic model. The output data of the electrochemical prediction sub-models in the model set, such as the lithium-ion intercalation / deintercalation model and the internal resistance prediction model (i.e., real-time lithium-ion concentration distribution and dynamic internal resistance values), are mapped to the electrochemical reaction input parameters of the battery basic model, enabling the basic model to simulate the charge transfer process inside the battery at low temperatures. Simultaneously, the calculation results of the thermal conduction prediction sub-models, such as the temperature field distribution model and the heat loss calculation model (e.g., real-time temperature of various areas of the cell and heat flux density of the heat dissipation channel), are integrated into the thermal analysis module of the basic model. This synchronously recreates the heat generation, transfer, and dissipation processes of the battery under low-temperature conditions, achieving coupled simulation of electrochemistry and thermal conduction. Subsequently, low-temperature operating parameters matching actual application scenarios are set, such as the ambient temperature gradient from -10℃ to -30℃ and different discharge rate requirements. Low-temperature twin simulation is then initiated. Through real-time model calculations and data feedback, the operating characteristics of the battery at low temperatures, such as charge / discharge voltage changes, capacity decay, and temperature distribution, are dynamically replicated. Finally, a battery low-temperature operating twin that accurately maps the low-temperature performance of the physical battery and supports subsequent performance prediction and repair strategy verification is generated.

[0036] In one possible implementation, step S220 further includes: Step S221: Perform outlier cleaning and standardization on the battery historical low-temperature operation dataset to obtain a usable battery low-temperature operation dataset.

[0037] Step S222: Match and classify each prediction task in the battery low temperature prediction task list with the available battery low temperature operation dataset to obtain the battery low temperature prediction task operation sample set.

[0038] Step S223: Select the battery prediction task model structure set according to the battery low temperature prediction task list.

[0039] Step S224: Based on the battery low temperature prediction task running sample set, perform prediction task training on the battery prediction task model structure set to generate the battery low temperature prediction task model set.

[0040] Specifically, for the historical low-temperature operation dataset of batteries collected by BMS, which includes data such as voltage, current, temperature, and capacity at different low temperatures, the 3σ principle is used to identify and remove invalid outliers such as voltage jumps and abnormal temperature fluctuations caused by sensor faults. Then, through Min-Max standardization, parameters of different magnitudes such as current, voltage, and temperature are uniformly mapped to a standard range to eliminate the impact of dimensional differences on subsequent training, and finally a usable low-temperature operation dataset of batteries with qualified data and uniform format is obtained.

[0041] Based on the electrochemical predictions in the battery low-temperature prediction task list, such as lithium-ion intercalation / deintercalation rate, internal resistance change and thermal conduction prediction, such as temperature field distribution and heat loss task classification, as well as the core data requirements of each sub-task, the available battery low-temperature operation datasets are matched, split and labeled. For example, voltage curves and capacity decay data are associated with the lithium-ion intercalation / deintercalation rate prediction task, and temperature data and heat dissipation power data of each area of ​​the cell are associated with the temperature field distribution prediction task. At the same time, task type labels and data source labels are added to each type of task sample to form a clearly classified battery low-temperature prediction task operation sample set that can be directly used for training.

[0042] Based on the core differences between the battery electrochemical prediction task type and the battery thermal conduction prediction task type in the battery low-temperature prediction task list, and the technical characteristics of each subtask under the two types of tasks, the model structure is selected as follows: For the subtasks under the electrochemical prediction task that need to capture the time-series change patterns, such as lithium-ion intercalation / deintercalation rate prediction and capacity decay trend prediction, it is necessary to analyze the data of battery charging and discharging voltage and current changing over time. The Long Short-Term Memory Network (LSTM) is selected as the basic model structure to accurately mine the dynamic correlations in the time-series data. For the subtasks under the thermal conduction prediction task that need to analyze spatial distribution features, such as cell temperature field distribution prediction and heat flux density prediction of heat dissipation channels, since it is necessary to process the spatial distribution information of temperature in different areas of the battery, a Convolutional Neural Network (CNN) combined with a finite element analysis model is selected. The spatial feature extraction capability of CNN and the physical field calculation advantage of the finite element model are used to improve the spatial prediction accuracy. For the simple static parameter prediction subtasks in both types of tasks, such as low-temperature static internal resistance prediction, traditional machine learning models such as gradient boosting trees or random forests are selected to improve training efficiency while ensuring prediction accuracy. Finally, these model structures adapted to different subtasks are integrated to form a set of battery prediction task model structures covering all prediction tasks.

[0043] The battery low-temperature prediction task sample set, including electrochemical prediction samples and thermal conduction prediction samples, is split into training, validation, and test sets according to a preset ratio, such as 7:2:1, to ensure that the data distribution of each subset matches the task type. Then, for different models in the battery prediction task model structure set, such as the LSTM model for the electrochemical task, the CNN-finite element model for the thermal conduction task, and the XGBoost model for static parameter prediction, targeted training is conducted using the corresponding task's training set data. Taking the electrochemical sub-task as an example, the LSTM model is trained using voltage / current time-series data from the training set, and the Adam optimizer is used to minimize the mean squared error (MSE) between the predicted and actual values. During training, the model performance is evaluated using the validation set after each iteration, and the mean absolute error (MAE) is calculated. Hyperparameters, such as the learning rate, number of iterations, and number of network layer nodes, are dynamically adjusted based on the validation results to avoid model overfitting. After all models are trained, the generalization ability is verified using the test set, and models that meet the preset accuracy threshold are selected. Finally, these models that are adapted to different low-temperature prediction tasks and meet the performance standards are integrated to generate a battery low-temperature prediction task model set covering all prediction needs of electrochemistry and thermal conduction.

[0044] In one possible implementation, step S222 further includes: Step S2221: Match and classify each prediction task in the battery low temperature prediction task list with the available battery low temperature operation dataset to obtain the battery low temperature prediction task associated dataset.

[0045] Step S2222: Determine the data association depth for the prediction task based on the target requirements of the battery low temperature prediction task.

[0046] Step S2223: According to the association depth of the prediction task data, perform association filtering and identification on the battery low temperature prediction task associated dataset to obtain the battery low temperature prediction task running sample set.

[0047] Specifically, the key data dimensions of each sub-task in the battery low-temperature prediction task list are first identified. For example, lithium-ion intercalation / deintercalation rate prediction requires matching real-time current and voltage change curves and electrode reaction time data at different low temperatures from the available battery low-temperature operation dataset; internal resistance change trend prediction requires linking charge / discharge cycle counts, cell temperature fluctuation data, and internal resistance monitoring values; temperature field distribution prediction requires matching heat dissipation channel flow rate, ambient temperature gradient, and data collected by temperature sensors in various areas of the cell; and heat loss prediction requires corresponding data such as heat dissipation power and insulation layer thickness. Then, according to the above data dimension requirements, the available battery low-temperature operation dataset is filtered and classified item by item. Data of the same sub-task type are grouped together, and each group is labeled with a corresponding task type, such as electrochemistry-intercalation / deintercalation rate or thermal conduction-temperature field. This ultimately forms a battery low-temperature prediction task-related dataset that corresponds one-to-one with each prediction task, ensuring that each task can obtain the core related data to support its subsequent analysis.

[0048] Based on the differences in prediction objectives among the sub-tasks in the battery low-temperature prediction task list, the data association depth is determined accordingly. For tasks aiming to predict short-term dynamic characteristics, such as predicting instantaneous charge-discharge voltage fluctuations at low temperatures and predicting thermal response speed during startup, the data association depth is defined as shallow association, requiring coverage of only the last 1-3 charge-discharge cycles and a single low-temperature condition, such as core parameters at typical temperatures in the target scenario, i.e., voltage, current, and real-time temperature data. For tasks aiming to predict long-term performance trends, such as predicting the low-temperature capacity decay rate after 100 cycles and predicting the internal resistance change trend under different low-temperature gradients, the data association depth is defined as medium association, requiring coverage of the last 50-100 cycles. For each charge / discharge cycle and 3-5 key low-temperature operating conditions, such as -10℃, -20℃, and -30℃, all-dimensional parameters are required, including capacity, internal resistance, cycle count, and temperature distribution data in each region. For tasks aimed at multi-factor coupled risk prediction, such as lithium plating risk prediction under low-temperature high-rate discharge and thermal runaway risk prediction under complex environmental temperature fluctuations, the data correlation depth is defined as deep correlation. It needs to cover different low-temperature operating conditions, different discharge rates, and different aging degrees, such as historical operating data under SOH of 80%~100%. At the same time, it should be correlated with battery structural parameters, such as tab layout, heat dissipation channel size, and material characteristic parameters, such as electrolyte ionic conductivity, to ensure that the data can support multi-variable coupled analysis and meet the task's accurate prediction requirements.

[0049] For the associated datasets corresponding to different prediction tasks, screening is carried out according to the established shallow, medium, and deep association depth standards. For short-term dynamic characteristic prediction tasks requiring shallow association, such as instantaneous voltage fluctuation prediction, core parameter data such as voltage, current, and real-time temperature under the last 1-3 charge-discharge cycles and single target low-temperature conditions are selected from the associated dataset, and redundant historical data exceeding the cycle range are removed. For long-term performance trend prediction tasks requiring medium association, such as cycle capacity decay prediction, full-dimensional data such as capacity, internal resistance, and cycle count under 3-5 key low-temperature gradients are extracted from the corresponding associated dataset to ensure coverage of the time span and operating condition range required by the task. For multi-factor coupled risk prediction tasks requiring deep association, such as low-temperature high-rate lithium plating risk prediction, complete operating data covering different low temperatures, different discharge rates, and different aging degrees are selected from the associated dataset, while retaining the battery structure and material characteristic correlation parameters and removing irrelevant operating conditions, such as room temperature data and duplicate records. After screening, corresponding association depth labels are added to the sample data of each type of task, such as shallow-short-term dynamics, deep-multi-factor coupling, and task type identification, such as electrochemistry-intercalation / deintercalation rate and thermal conduction-temperature field. Finally, the data are integrated to form a battery low-temperature prediction task running sample set with clear classification and data dimensions that are highly matched with task requirements.

[0050] In one possible implementation, step S300 further includes: Step S310: Based on the battery low-temperature performance evaluation index, obtain the battery low-temperature performance dimension index, which includes electrochemical performance dimension index, thermal management performance dimension index, and safety performance dimension index.

[0051] Step S320: Perform hierarchical index splitting and weight allocation on the battery low-temperature performance dimension index to obtain the battery dimension sub-level index set and sub-level index weight factor set.

[0052] Step S330: Perform battery performance prediction on the battery low-temperature operation twin according to the battery dimension sub-level index set and sub-level index weight factor set, and generate the battery low-temperature operation prediction performance parameters.

[0053] Specifically, guided by the assessment needs covering the entire lifecycle performance of batteries operating at low temperatures, the evaluation indicators for battery low-temperature performance are categorized and broken down into three core dimensions: electrical performance, thermal management effectiveness, and safety risk control. These categorized indicators result in the following battery low-temperature performance dimensions: Electrochemical performance dimensions focus on core electrical performance parameters of the battery at low temperatures, including low-temperature capacity retention (e.g., the ratio of actual discharge capacity to room temperature capacity at -20℃ and -30℃), low-temperature charge / discharge efficiency (e.g., charge acceptance and discharge energy output efficiency in different low-temperature ranges), and voltage plateau stability (e.g., voltage drop amplitude and duration of stability during low-temperature discharge), used to quantitatively assess the battery's energy storage and release capabilities at low temperatures. Thermal management performance dimensions revolve around the battery's heat transfer and temperature control effectiveness at low temperatures, including temperature distribution uniformity (e.g., the maximum temperature difference between the cell interior and surface) and low-temperature start-up. The heating rate, such as the time it takes to rise from ambient low temperature to a suitable operating temperature, and the heat transfer efficiency of the heat dissipation channel, such as the amount of heat transferred through the heat dissipation structure per unit time, are used to ensure the thermal stability of the battery during low-temperature operation. The safety performance dimension focuses on preventing potential safety risks under low-temperature conditions, including the risk level of lithium plating during low-temperature charging, such as the lithium plating threshold and probability of occurrence; the internal resistance growth rate, such as the change in internal resistance after long-term low-temperature cycling; and the stability of low-temperature mechanical properties, such as the shell's resistance to cracking and the integrity of the electrode structure. These three dimensions together constitute the system's battery low-temperature performance evaluation system.

[0054] The three dimensions of battery low-temperature performance are hierarchically broken down, transforming abstract dimensions into specific, calculable sub-level indicators. For electrochemical performance, the indicators are broken down into sub-indicators such as "capacity retention rate at -10℃", "capacity retention rate at -20℃", "charge and discharge efficiency at -30℃", and "low-temperature voltage plateau voltage drop". For thermal management performance, the indicators are broken down into sub-indicators such as "maximum temperature difference on cell surface", "maximum temperature difference inside cell", "temperature rise value after 10 minutes of low-temperature start-up", and "heat loss per unit time in heat dissipation channels". For safety performance, the indicators are broken down into sub-indicators such as "lithium plating amount at 0.5C low-temperature charging", "internal resistance growth rate at 1C low-temperature discharge", and "shell deformation rate after standing at -30℃". These are integrated to form a comprehensive set of battery dimension sub-level indicators. Subsequently, considering the low-temperature application requirements of the target battery, such as prioritizing start-up heating rate and safety in automotive scenarios and prioritizing capacity retention in energy storage scenarios, the Analytic Hierarchy Process (AHP) was used to assign weight factors to each sub-level indicator. For example, in automotive scenarios, the weight of "temperature rise value in 10 minutes of low-temperature start-up" was set to 0.22, and the weight of "lithium plating amount at 0.5C low-temperature charging" was set to 0.18; in energy storage scenarios, the weight of "capacity retention rate at -20℃" was set to 0.25, and the weight of "charge and discharge efficiency at -30℃" was set to 0.2. This resulted in a set of sub-level indicator weight factors adapted to the scenario requirements, ensuring that the evaluation results accurately reflect the core performance priorities.

[0055] First, based on the classification logic of the battery dimension sub-level index set, corresponding data acquisition and calculation modules are set in the battery low-temperature operation twin. For electrochemical sub-indicators, such as capacity retention rate at -10℃ / -20℃ and low-temperature charge-discharge efficiency, the twin simulates charge-discharge cycles under different low-temperature conditions, collects the ratio of discharge capacity and charging energy to the room temperature reference value in real time, and calculates the capacity retention rate and efficiency value for each temperature range. For thermal management sub-indicators, such as cell temperature difference and low-temperature start-up temperature rise value, the twin's thermal field simulation function monitors the temperature distribution data of the cell surface / internal area, calculates the maximum temperature difference and the temperature rise after a specified time after start-up, such as 10 minutes. For safety sub-indicators, such as lithium plating amount and internal resistance growth rate, the twin's electrochemical reaction simulation module analyzes the efficiency of lithium ion insertion into the negative electrode during low-temperature charging to quantify lithium plating amount, and compares the internal resistance data before and after multiple cycles to calculate the growth rate. Subsequently, the calculation results of each sub-indicator are weighted and integrated by combining the sub-level indicator weight factor set. For example, in the vehicle scenario, the temperature rise value after 10 minutes of low temperature start-up (weight 0.22) and the capacity retention rate at -20℃ (weight 0.18) are converted into a comprehensive score according to the weight ratio. Finally, a battery low-temperature operation prediction performance parameter is generated, which includes the original calculated value of each sub-indicator, the weighted score, and the performance level judgment, so as to comprehensively and quantitatively reflect the battery's comprehensive performance in low-temperature environment.

[0056] In one possible implementation, step S400 further includes: Step S410: Analyze the repair strategy for the predicted low-temperature operation performance parameters of the battery to obtain the threshold values ​​for the low-temperature performance repair strategy parameters of the battery.

[0057] Step S420: Randomly select multiple battery repair strategy parameters within the threshold of the battery low-temperature performance repair strategy parameters, and use the battery low-temperature operation twin to perform repair simulation evaluation on the multiple battery repair strategy parameters to obtain multiple battery low-temperature performance repair effects.

[0058] Step S430: Based on the low-temperature performance repair effects of the multiple batteries, perform global optimization on the parameters of the multiple battery repair strategies to construct a battery low-temperature performance repair strategy solution library.

[0059] Specifically, we identified the non-compliant performance parameters in the battery's low-temperature operation prediction, such as a capacity retention rate of only 68% at -20℃, a temperature rise of less than 4℃ in 10 minutes after low-temperature start-up, and lithium plating exceeding 5mg during 0.5C charging. We then categorized and matched repair strategies according to three main directions: electrochemical performance repair, thermal management performance repair, and safety performance repair. For insufficient capacity retention, we associated strategies such as electrolyte temperature control and negative electrode interface modification. For slow temperature rise, we associated strategies such as insulation layer power adjustment and cell preheating control. For lithium plating risk, we associated strategies such as charging current gradient limitation and charging cutoff voltage optimization. Subsequently, considering the material characteristics of lithium iron phosphate batteries, such as the low-temperature ionic conductivity threshold of the electrolyte, the thermal stability range of electrode materials, and engineering safety boundaries, such as the upper limit of casing temperature resistance and the power carrying capacity limit of the circuit, the reasonable value ranges of key parameters for each repair strategy were analyzed. For example, electrolyte temperature control needs to avoid freezing at low temperatures (below -10℃) and deterioration at high temperatures (above 15℃), with a threshold defined as -5℃ to 10℃; insulation layer power adjustment needs to balance energy consumption and heating efficiency, with a threshold defined as 5W to 18W based on the upper limit of battery energy consumption; charging current limitation needs to avoid excessively low current leading to excessively long charging time (below 0.1C) and excessive current accelerating lithium plating (above 0.6C), with a threshold defined as 0.1C to 0.5C. Finally, the parameter ranges of all repair strategies were integrated to form a clearly categorized set of battery low-temperature performance repair strategy parameter thresholds that are both effective and safe.

[0060] Based on the threshold parameters of the battery low-temperature performance repair strategy, such as electrolyte temperature -5℃~10℃, heat preservation power 5W~18W, and charging current 0.1C~0.5C, multiple sets of differentiated battery repair strategy parameters are generated using a uniform random sampling method. Each set of parameters must cover all repair dimensions. For example, the first set is electrolyte temperature 8℃, heat preservation power 15W, and charging current 0.4C; the second set is electrolyte temperature 3℃, heat preservation power 10W, and charging current 0.2C; and the third set is electrolyte temperature -2℃, heat preservation power 7W, and charging current 0.3C, etc., to ensure that the parameter combinations cover different value ranges within the threshold and avoid sample bias. Subsequently, each set of parameters was input into the battery low-temperature operating twin, and low-temperature operating conditions matching actual applications were set, such as -20℃ environment and 1C discharge requirement, to simulate the battery operation process after the implementation of the repair strategy: the twin collected and calculated changes in key performance indicators in real time, such as capacity retention rate increasing from 68% to 79%, heating rate increasing from 4℃ / 10min to 9℃ / 10min, and lithium plating decreasing from 5mg to 2mg, forming "before-repair" performance comparison data for each set of parameters. Finally, multiple battery low-temperature performance repair effects corresponding to all repair strategy parameters were obtained, comprehensively reflecting the repair efficiency of different parameter combinations.

[0061] The evaluation standard is based on the comprehensive performance of multiple battery low-temperature performance repair methods. It considers the core requirements of low-temperature application scenarios, such as prioritizing heating rate and safety risks in automotive scenarios, and focusing on capacity recovery rate and energy consumption control in energy storage scenarios. Indicators such as capacity retention improvement, heating efficiency, lithium plating reduction, and energy cost are weighted and converted into a comprehensive score to quantitatively evaluate the effectiveness of each repair strategy parameter set. Subsequently, a genetic algorithm is used for global optimization: the top 30% of parameter sets with the comprehensive score are used as initial parent samples. Through crossover, such as combining high-capacity recovery parameters with high-heating efficiency parameters, and mutation, such as fine-tuning electrolyte temperature or charging current values, a new generation of repair parameters is generated. These parameters are then input into a battery low-temperature operating twin to verify the repair effect. After 5-8 rounds of iteration, parameter combinations that perform well under different low-temperature conditions, such as -10℃, -20℃, and -30℃, are gradually selected. At the same time, special solutions for specific needs are retained, such as rapid heating priority solutions and low-energy consumption repair solutions. Finally, all optimized parameter combinations were categorized and organized: modules were divided according to the low-temperature operating conditions and core requirements, such as the -20℃ vehicle scenario - rapid start solution and the -30℃ energy storage scenario - capacity recovery solution. The repair parameter values, expected repair effects, applicable scenario boundaries and operation precautions of each solution were clarified. In the end, a battery low-temperature performance repair strategy solution library with a clear structure, strong adaptability and can directly guide actual operation and maintenance was constructed.

[0062] In one possible implementation, step S430 further includes: Step S431: Optimize the multiple battery repair strategy parameters according to the multiple battery low-temperature performance repair effects to obtain N parent repair strategy parameters.

[0063] Step S432: Based on the N parent repair strategy parameters, perform extended learning and update to generate N battery repair strategy parameter clusters.

[0064] Step S433: The total repair effect of the N battery repair strategy parameter clusters is evaluated by the battery low-temperature operation twin, and the parameter cluster with the largest total repair effect is used as the battery low-temperature performance repair strategy scheme library.

[0065] Specifically, based on the comprehensive performance of multiple battery low-temperature performance repairs, and combined with the key requirements of low-temperature application scenarios, such as the heating rate in automotive scenarios and the capacity recovery rate in energy storage scenarios, the comprehensive score is calculated according to preset weights for indicators such as capacity improvement, reduction of lithium plating risk, and energy consumption cost. The top N parameters with the highest comprehensive scores are selected from all battery repair strategy parameters, such as N=20. These parameters have outstanding repair efficiency and are suitable for batteries with different basic performance shortcomings, and are used as the N parent repair strategy parameters for subsequent optimization.

[0066] Next, for each parent repair strategy parameter, an expansion and learning update method is adopted, which involves fine-tuning local parameters and adapting to different operating conditions. In key dimensions of the parent parameters, such as electrolyte temperature, heat preservation power, and charging current, small adjustments are made at 5% to 10% of the repair strategy parameter threshold to generate 5 to 8 differentiated sub-parameters. At the same time, combined with different low-temperature operating conditions, such as -10℃, -20℃, and -30℃, and battery aging degree, such as SOH 80% to 100%, parameter variants adapted to multiple scenarios are expanded. Finally, each parent parameter corresponds to a set containing 20 to 30 related parameters, that is, N battery repair strategy parameter clusters, ensuring that the parameters in each cluster retain the efficient repair characteristics of the parent while covering a wider range of application scenarios.

[0067] A unified evaluation standard for the total repair effect is established for N battery repair strategy parameter clusters. With full-condition coverage and full-state adaptability as the core, it covers three key low-temperature conditions (-10℃, -20℃, -30℃) and three typical aging states (80%, 90%, 100% battery health). For all parameters within each parameter cluster, a battery low-temperature operating twin is input one by one to simulate the repair process under different conditions and states. Key indicators such as capacity retention improvement, temperature rise rate improvement, and lithium plating reduction are collected and calculated. Then, based on scenario priority weights (e.g., 0.3 weight for temperature rise rate in automotive scenarios and 0.3 weight for capacity recovery in energy storage scenarios), a single-scenario repair score for each parameter is calculated, and these scores are summed to obtain the total score for the parameter across all scenarios. Finally, the total scores for all parameters within each parameter cluster are summarized to obtain the total repair effect of each cluster. A higher total score indicates stronger adaptability of the cluster to different low-temperature environments and batteries with different aging levels, and better overall repair stability and universality. Finally, the parameter clusters with the largest total repair effect were selected, and the parameters within the clusters were classified and sorted according to low temperature conditions, battery aging status, and core repair targets, such as -20℃-SOH80%-capacity-priority repair and -30℃-SOH90%-safety-priority repair. The specific values, operation steps, expected repair effects, and applicable boundaries of each parameter were clarified, ultimately forming a battery low temperature performance repair strategy solution library that covers multiple scenarios, has strong adaptability, and can be directly implemented.

[0068] In one possible implementation, step S432 further includes: Step S4321: Based on the N parent repair strategy parameters, perform crossover mutation expansion to obtain N repair strategy parameter populations.

[0069] Step S4322: Perform parameter learning and updating in the N repair strategy parameter populations respectively until the preset convergence requirement is met, and generate the N battery repair strategy parameter clusters.

[0070] Specifically, for the N parent repair strategy parameters, such as high-efficiency parameters covering core dimensions like electrolyte temperature, heat preservation power, and charging current, crossover and mutation operations from a genetic algorithm are used for expansion: In the crossover stage, two parent parameters are randomly selected as parents, and their key repair dimensions, such as electrolyte temperature range and charging current gradient, are fragmented and exchanged to generate offspring parameters that combine the advantages of both parents. For example, the "electrolyte temperature 5℃" of parent A is combined with the "charging current 0.3C" of parent B. In the mutation stage, according to a preset mutation probability, such as 5%~8%, the dimensional values ​​of some offspring parameters are slightly adjusted. For example, the heat preservation power is slightly adjusted from 12W to 11W or 13W to ensure parameter diversity. Through multiple rounds of crossover and mutation, each parent parameter generates a set containing 50~80 differentiated parameters, thus forming N repair strategy parameter populations, effectively expanding the parameter coverage.

[0071] For each repair strategy parameter population, containing 50-80 differentiated repair parameters, a simulation evaluation-feedback optimization cyclical mechanism is adopted for parameter learning and updating: First, all parameters in the population are input one by one into the battery low-temperature operation twin. The repair process is simulated under full low-temperature conditions, i.e. -10℃, -20℃, -30℃ and full battery aging state, i.e. SOH 80%~100%. The weighted comprehensive score of capacity improvement, heating efficiency, lithium plating risk reduction rate and energy consumption cost is used as the evaluation index to select the top 40% of high-quality parameters, while eliminating the bottom 30% of inefficient parameters. For the retained high-quality parameters, small fine-tuning is performed on their key dimensions, such as electrolyte temperature and charging current, at 3%~5% of the repair strategy parameter threshold to generate new supplementary parameters and maintain the stability of the population size. After each round of evaluation, screening, and fine-tuning, the average comprehensive score of all parameters within the population is calculated. If the fluctuation range of the average score for three consecutive rounds is less than the preset convergence threshold, such as 1.5%, the population is considered to have met the convergence requirements, indicating that the parameters within the population have stable and excellent repair performance. After learning and updating N parameter populations according to this process, each converged population is organized into a set with a clear structure and unified performance, ultimately generating N battery repair strategy parameter clusters.

[0072] In one possible implementation, step S400 further includes: Step S440: Based on the battery low-temperature performance repair strategy scheme library, determine the target battery low-temperature performance repair strategy scheme with the best repair effect.

[0073] Step S450: If the actual performance repair effect of the target battery low temperature performance repair strategy does not reach the preset battery low temperature performance, then the battery operation control is performed based on the suboptimal battery low temperature performance repair strategy in the battery low temperature performance repair strategy library.

[0074] Specifically, the actual state and operating conditions of the current battery are analyzed, including the ambient temperature, battery health, core performance shortcomings, and application scenario requirements. Based on these conditions, the corresponding classification module is located in the battery low-temperature performance repair strategy library. Key information of all repair solutions under this module is extracted, including expected repair effects (capacity increase, heating efficiency, reduction in lithium plating), and implementation costs (energy consumption, operational complexity, and applicability). A comprehensive evaluation model is established, prioritizing expected effects and implementation costs according to scenario (e.g., heating efficiency weighted at 0.3 for automotive scenarios and capacity recovery weighted at 0.3 for energy storage scenarios), converting these into quantitative scores. The scores of all solutions are compared, and the solution with the highest score that perfectly matches the battery's actual state and requirements is selected as the target battery low-temperature performance repair strategy with the best repair effect. Its specific execution parameters are also defined, such as the electrolyte temperature control range, insulation power setting, charging current gradient, and operating procedure.

[0075] After applying the target battery low-temperature performance repair strategy to the actual battery operation control, the battery management system (BMS) collects and monitors the actual performance repair data of the battery in real time, including but not limited to indicators such as low-temperature capacity retention rate, cell heating rate, lithium plating monitoring value, and internal resistance change. These actual data are then compared with preset battery low-temperature performance standards, such as capacity retention rate ≥80% at -20℃, temperature rise ≥6℃ after 10 minutes of startup, and lithium plating ≤3mg. If the monitoring finds that the actual performance repair effect does not meet the preset standards, for example, the actual capacity retention rate is only 75% or the heating rate is only 4.8℃ / 10 minutes, then it is determined that the current target solution is not well adapted to the actual battery operating state. At this point, the suboptimal repair strategy solution matching the current operating condition is immediately retrieved from the battery low-temperature performance repair strategy solution library. This solution has a comprehensive score second only to the target solution in the solution library ranking, and has verified effective repair records in similar performance bottleneck scenarios. The original target solution is quickly replaced to execute battery operation control, while the actual repair effect of the suboptimal solution is continuously tracked until the battery performance reaches the preset standard, ensuring that the battery always maintains a stable and safe operating state in low-temperature environments.

[0076] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0077] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0078] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for repairing the low-temperature performance of lithium iron phosphate batteries, characterized in that, The method includes: Obtain low-temperature application scenario information for the target lithium iron phosphate battery, and perform dynamic optimization of the battery based on the low-temperature application scenario information to obtain a battery structure optimization parameter system; By collecting historical low-temperature operation data of the battery through BMS, low-temperature twin simulation is performed based on the battery structure optimization parameter system and the historical low-temperature operation data of the battery to generate a low-temperature operation twin of the battery. Construct a battery low-temperature performance evaluation index, predict the battery performance of the battery low-temperature operation twin according to the battery low-temperature performance evaluation index, and generate battery low-temperature operation prediction performance parameters. The battery's predicted low-temperature operation performance parameters are analyzed for repair strategies, a battery low-temperature performance repair strategy library is constructed, and battery operation control is performed based on the battery low-temperature performance repair strategy library.

2. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 1, characterized in that, The battery structure optimization parameter system was obtained, including: Construct a database of lithium iron phosphate battery formulations and a database of lithium iron phosphate battery structures; The lithium iron phosphate battery formula database and lithium iron phosphate battery structure database are matched and optimized with the low-temperature application scenario information to obtain battery formula optimization data and battery structure optimization data. Based on the battery formulation optimization data and battery structure optimization data, the formulation and structure data of the target lithium iron phosphate battery are dynamically optimized to obtain battery formulation optimization parameters and battery structure optimization parameters. Based on the battery formulation optimization parameters and battery structure optimization parameters, the battery structure optimization parameter system is obtained.

3. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 1, characterized in that, Generate a battery low-temperature operation twin, including: Construct a battery low-temperature prediction task list, which includes battery electrochemical prediction task types and battery thermal conduction prediction task types; According to the battery low temperature prediction task list, the battery historical low temperature operation dataset is classified and predicted to generate a battery low temperature prediction task model set. Based on the battery structure optimization parameter system, a three-dimensional model is constructed to build the basic battery model. The battery low-temperature prediction task model set is coupled to the battery basic model for low-temperature twin simulation to generate a battery low-temperature operation twin.

4. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 3, characterized in that, Generate a set of models for battery low-temperature prediction tasks, including: The battery historical low-temperature operation dataset is cleaned of outliers and standardized to obtain a usable battery low-temperature operation dataset; Each prediction task in the battery low temperature prediction task list is matched and classified with the available battery low temperature operation dataset to obtain a battery low temperature prediction task operation sample set. Based on the battery low-temperature prediction task list, select the battery prediction task model structure set; Based on the battery low-temperature prediction task running sample set, the battery prediction task model structure set is trained to generate the battery low-temperature prediction task model set.

5. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 4, characterized in that, Obtain the sample set for the battery low-temperature prediction task, including: Each prediction task in the battery low temperature prediction task list is matched and classified with the available battery low temperature operation dataset to obtain the battery low temperature prediction task associated dataset. Based on the requirements and objectives of the battery low-temperature prediction task, determine the data correlation depth for the prediction task; The battery low temperature prediction task associated dataset is associated and filtered according to the association depth of the prediction task data to obtain the battery low temperature prediction task running sample set.

6. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 1, characterized in that, Generate predicted performance parameters for battery low-temperature operation, including: Based on the battery low-temperature performance evaluation index, the battery low-temperature performance dimension index is obtained, which includes electrochemical performance dimension index, thermal management performance dimension index, and safety performance dimension index. The battery low-temperature performance dimension index is hierarchically split and weighted to obtain the battery dimension sub-level index set and sub-level index weight factor set; Battery performance is predicted for the battery low-temperature operation twin based on the battery dimension sub-level index set and sub-level index weight factor set, generating the battery low-temperature operation prediction performance parameters.

7. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 1, characterized in that, Construct a library of battery low-temperature performance repair strategies, including: The battery low-temperature operation prediction performance parameters are analyzed to obtain the battery low-temperature performance repair strategy parameter thresholds. Within the threshold of the battery low-temperature performance repair strategy parameters, multiple battery repair strategy parameters are randomly selected. The battery low-temperature operation twin is used to perform repair simulation evaluation on the multiple battery repair strategy parameters to obtain the low-temperature performance repair effects of multiple batteries. Based on the low-temperature performance repair effects of the multiple batteries, the parameters of the multiple battery repair strategies are globally optimized to construct a battery low-temperature performance repair strategy solution library.

8. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 7, characterized in that, Based on the low-temperature performance repair effects of the multiple batteries, the parameters of the multiple battery repair strategies are globally optimized to construct a battery low-temperature performance repair strategy solution library, including: Based on the low-temperature performance repair effect of the multiple batteries, the parameters of the multiple battery repair strategies are optimized to obtain N parent repair strategy parameters; Based on the aforementioned N parent repair strategy parameters, an extended learning update is performed to generate N battery repair strategy parameter clusters; The battery low-temperature operation twin is used to evaluate the total repair effect of the N battery repair strategy parameter clusters, and the parameter cluster with the largest total repair effect is used as the battery low-temperature performance repair strategy scheme library.

9. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 8, characterized in that, Generate N battery repair strategy parameter families, including: Based on the N parent repair strategy parameters, crossover and mutation expansion are performed to obtain N repair strategy parameter populations; Parameter learning and updating are performed in each of the N repair strategy parameter populations until the preset convergence requirement is met, thereby generating the N battery repair strategy parameter clusters.

10. The method for repairing the low-temperature performance of a lithium iron phosphate battery as described in claim 1, characterized in that, Battery operation control is performed based on the aforementioned battery low-temperature performance repair strategy library, including: Based on the battery low-temperature performance repair strategy library, determine the target battery low-temperature performance repair strategy with the best repair effect; If the actual performance repair effect of the target battery low-temperature performance repair strategy fails to reach the preset battery low-temperature performance, then the battery operation control will be performed based on the suboptimal battery low-temperature performance repair strategy in the battery low-temperature performance repair strategy library.

Images