An externally-mounted battery AI intelligent diagnosis and repair system

Abstract

This invention provides an external battery AI-powered intelligent diagnostic and repair system. It includes a multi-dimensional sensing module that collects battery data, utilizes deep reinforcement learning algorithms and spatiotemporal convolutional neural networks on an edge computing terminal to analyze battery status in real time, establishes a lithium plating kinetic model, and provides early warnings. A cloud server combines multiphysics coupling modeling and deep reinforcement learning to predict battery life and formulate repair strategies, constructing a digital twin platform. The intelligent repair module employs a quantum genetic algorithm-optimized energy routing device and pulse repair unit to achieve cell-level balancing and SEI film repair. A collaborative control module links the BMS, PCS, and EMS for system-level optimization management. This invention can extend battery life and reduce the total lifecycle cost and safety risks of energy storage systems.

Description

Technical Field

[0001] This invention relates to the field of battery cell management technology, and more specifically, to an external battery AI intelligent diagnostic and repair system. Background Technology

[0002] In the field of energy storage battery management, traditional battery management systems (BMS) mainly rely on threshold monitoring of macroscopic parameters such as voltage and temperature, as well as passive equalization technology. However, with the increasing demands of new power systems for energy storage safety, economy, and lifespan, traditional technologies have revealed significant shortcomings: Detection methods are outdated; existing BMSs can only identify abnormalities in macroscopic parameters at the cell level, such as overvoltage and overtemperature, and cannot capture early signs of microscopic damage such as lithium plating and SEI film thickening, leading to delayed thermal runaway warnings until the irreversible stage. Repair capabilities are insufficient; passive equalization technology dissipates energy through resistance, resulting in low efficiency and failing to suppress the discretization trend of capacity / internal resistance between cells, accelerating the lifespan degradation of battery clusters. Model accuracy is limited; traditional lifespan prediction models are based on cell-level macroscopic parameters, ignoring the effects of thermal-electric-chemical multi-field coupling and material interface phase transitions on performance degradation. Collaborative management is lacking; BMS, PCS, and EMS operate independently, lacking system-level repair strategy coordination, leading to high repair energy consumption and a high risk of secondary failures.

[0003] In implementing the embodiments of the present invention, the prior art has at least the following problems or defects: outdated detection methods, insufficient repair capabilities, limited model accuracy, and lack of collaborative management. These problems lead to increased costs throughout the entire life cycle of energy storage systems and prominent safety hazards. Summary of the Invention

[0004] This invention provides an AI-powered intelligent diagnostic and repair system and method for external batteries.

[0005] In a first aspect of the present invention, an external battery AI intelligent diagnostic and repair system is provided, comprising: The multi-dimensional sensing module is configured to collect data on the battery's current, voltage, temperature, gas, internal resistance, smoke, and pressure. The edge computing terminal integrates deep reinforcement learning algorithms and spatiotemporal convolutional neural networks to analyze battery dynamic parameters in real time, establish lithium plating kinetic models, and output latent degradation features and safety warnings. The cloud server is configured to generate battery lifetime prediction and adaptive equilibrium repair strategies through multi-physics coupled modeling and deep reinforcement learning, and to build a digital twin platform that integrates electrochemical-physical simulation. The intelligent repair module includes an intelligent bidirectional energy routing device optimized by a quantum genetic algorithm and a pulse repair unit; wherein the intelligent bidirectional energy routing device implements micro-operation balancing, and the pulse repair unit generates a microcurrent with a frequency of 1-10kHz to actively repair SEI film damage. The collaborative control module enables collaborative processing of heterogeneous data between edge computing terminals and cloud servers, and links the battery management system (BMS), power conversion system (PCS), and energy management system (EMS) for system-level thermal management, energy routing, and strategy optimization.

[0006] Furthermore, the pulse repair unit injects an asymmetric microcurrent waveform to optimize the SEI film structure when the AI ​​predicts that the amount of lithium plating exceeds the dynamic threshold or every 500 cycles. The intelligent bidirectional energy routing device, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through a multi-modal balanced topology, suppressing parameter discretization between cells.

[0007] Furthermore, the spatiotemporal convolutional neural network tracks the causal chain between material interface phase transitions and macroscopic performance degradation by decoupling the thermo-electro-chemical coupling effect; The digital twin platform synchronizes the evolution of microscopic parameters such as cell internal resistance and polarization characteristics with real-time monitoring data to generate precise repair instructions.

[0008] Furthermore, the collaborative control module triggers the pre-start of the fire protection system when repairing an anomaly, and switches to a low-rate charging and discharging mode during the repair period; The system supports cross-brand plug-and-play functionality, is compatible with multi-cluster parallel configurations, and requires no modification to the existing energy storage architecture.

[0009] Furthermore, the amplitude of the microcurrent output by the pulse repair unit is less than 0.1C, and the duration is 10%-30% of the battery resting period.

[0010] In a second aspect of the present invention, an AI-powered intelligent diagnostic and repair method for an external battery is provided, comprising: Real-time collection of multi-dimensional battery data, analysis of hidden failure factors through edge computing terminals, and output of safety warnings; The cloud server integrates historical data with digital twin models to predict lifespan degradation trajectories and formulate repair strategies; When the amount of lithium plating exceeds the dynamic threshold or reaches 500 cycles, the pulse repair unit and intelligent bidirectional energy routing device are activated to simultaneously implement SEI film optimization and cell-level balancing. Based on the dynamic optimization strategy of repair effect, the algorithm of edge computing terminal and cloud server can be autonomously iterated and system-level collaboratively managed.

[0011] Furthermore, during the safety intervention window of the T1-T2 phase, priority should be given to repairing cells with internal resistance dispersion > 5% or capacity decay rate > 3%. After the repair is completed, the recovery degree of cell polarization characteristics is verified by digital twin model. If the threshold is not reached, a second repair is triggered.

[0012] Furthermore, the intelligent bidirectional energy routing device dynamically optimizes the energy transfer path through a quantum genetic algorithm, adjusting the energy distribution between battery cells with a response time of up to seconds.

[0013] Furthermore, the system-level collaborative management includes dynamic adjustment of liquid cooling flow, cross-module energy routing compensation, and optimization of the energy management system (EMS) charging and discharging strategy; All repair operations are conducted on the premise of not affecting the grid-connected operation of the energy storage system, and the energy consumption of repair is less than 0.5% of the total system capacity.

[0014] Furthermore, the energy transfer path optimized by the quantum genetic algorithm reduces the inter-cell capacity standard deviation to below 1.23%, voltage dispersion to below 45%, internal resistance dispersion to below 38%, and temperature dispersion to below 51%.

[0015] The embodiments of the present invention have at least the following beneficial effects: 1. By collecting battery current, voltage, temperature, gas, internal resistance, smoke, and pressure data through a multi-dimensional sensing module, and combining the deep reinforcement learning algorithm and spatiotemporal convolutional neural network integrated into the edge computing terminal, the battery dynamic parameters can be analyzed in real time and a lithium plating kinetic model can be established. This allows for the early detection of micro-damage such as lithium plating, effectively solving the problem of lagging traditional technical detection methods. It realizes the transformation from macro-parameter monitoring to early warning of micro-damage, greatly improving the safety and reliability of battery management, advancing the safety intervention window to the T1-T2 stage, and avoiding safety accidents such as thermal runaway.

[0016] 2. The intelligent repair module includes an intelligent bidirectional energy routing device optimized by a quantum genetic algorithm and a pulse repair unit. The former, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through multimodal equalization topology to suppress parameter discretization between cells. The latter can generate kHz-level microcurrents to actively repair SEI film damage. This active repair method overcomes the problems of low efficiency and inability to suppress parameter discretization between cells in traditional passive equalization technology. It achieves simultaneous cell-level micro-repair and system-level thermal-electric synergistic management, effectively extending battery life, reducing the total life cycle cost of the energy storage system, and improving the economics of battery management.

[0017] 3. The system adopts a cloud-edge collaborative architecture, integrating a multi-physics digital twin model. The cloud server generates battery life prediction and adaptive equalization repair strategies through multi-physics coupled modeling and deep reinforcement learning, and constructs a digital twin platform integrating electrochemical-physical simulation. This enables heterogeneous data collaborative processing between edge computing terminals and the cloud server, and links with BMS, PCS, and EMS for system-level thermal management, energy routing, and strategy optimization. This overcomes the bottlenecks of traditional technical model accuracy limitations and lack of collaborative management, improves the accuracy of battery life prediction, optimizes the management strategy of the entire energy storage system, reduces repair energy consumption, reduces the occurrence of secondary failures, and improves the overall performance and safety of the energy storage system. Attached Figure Description

[0018] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein: Figure 1 This is a schematic diagram of the structure of an external battery AI intelligent diagnostic and repair system provided in an embodiment of the present invention; Figure 2 This is a graph showing the change in lithium plating before and after the repair. Figure 3 It is a visual comparison of the dynamic balancing effect of the battery pack; Figure 4 This is a data chart for predicting and warning of thermal runaway stages. Detailed Implementation

[0019] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make the invention more thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

[0020] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0021] It should be noted that the number of any elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0022] The following is for reference. Figure 1 , Figure 1This is a schematic diagram of the structure of an external battery AI intelligent diagnostic and repair system provided in an embodiment of the present invention. Figure 1 As shown, an external battery AI intelligent diagnostic and repair system includes: The multi-dimensional sensing module 101 is configured to collect data on the battery's current, voltage, temperature, gas, internal resistance, smoke, and pressure. Edge computing terminal 102 integrates deep reinforcement learning algorithms and spatiotemporal convolutional neural networks to analyze battery dynamic parameters in real time, establish lithium plating kinetic models, and output latent degradation features and safety warnings. Cloud server 103 is configured to generate battery life prediction and adaptive equilibrium repair strategies through multi-physics coupled modeling and deep reinforcement learning, and to build a digital twin platform that integrates electrochemical-physical simulation. The intelligent repair module 104 includes an intelligent bidirectional energy routing device optimized by a quantum genetic algorithm and a pulse repair unit; wherein the intelligent bidirectional energy routing device implements micro-operation balancing, and the pulse repair unit generates a microcurrent with a frequency of 1-10kHz to actively repair SEI film damage. The collaborative control module 105 enables collaborative processing of heterogeneous data between edge computing terminals and cloud servers, and links the battery management system (BMS), power conversion system (PCS), and energy management system (EMS) for system-level thermal management, energy routing, and strategy optimization.

[0023] It should be noted that the external battery AI intelligent diagnostic and repair system of this invention is an innovative battery management system. Its core lies in the collaborative work of a multi-dimensional sensing module, an edge computing terminal, a cloud server, an intelligent repair module, and a collaborative control module to achieve precise monitoring, diagnosis, and repair of the battery's state. The multi-dimensional sensing module is responsible for collecting various key battery data, such as current, voltage, and temperature. These data directly reflect the battery's operating state and are crucial for subsequent diagnosis and repair. The edge computing terminal uses deep reinforcement learning algorithms and spatiotemporal convolutional neural networks to analyze the collected data in real time, establishing a lithium plating kinetic model. This is a physical model describing the lithium plating phenomenon inside the battery, which can predict microscopic changes inside the battery, thereby identifying potential fault risks in advance. The cloud server generates battery life prediction and adaptive equilibrium repair strategies through multi-physics coupling modeling and deep reinforcement learning, and constructs a digital twin platform integrating electrochemical-physical simulation. The digital twin platform is a virtual model that can reflect the battery's physical state in real time, providing a basis for precise repair. The intelligent repair module includes a quantum genetic algorithm-optimized intelligent bidirectional energy routing device and a pulse repair unit. These devices can perform targeted repair operations based on diagnostic results, such as micro-manipulation balancing and SEI membrane repair. The collaborative control module ensures coordinated operation among all parts of the system, achieving system-level optimized management.

[0024] Specifically, the battery data collected by the multi-dimensional sensing module includes current, voltage, temperature, gas, internal resistance, smoke, and pressure. These data reflect the battery's electrochemical reaction intensity, energy output level, internal thermal state, potential chemical reaction products, internal resistance changes, potential combustion risk, and internal pressure. Deep reinforcement learning algorithms in edge computing terminals are machine learning methods that learn optimal strategies through trial and error. They can dynamically adjust model parameters based on real-time data collection to better adapt to changes in battery state. Spatiotemporal convolutional neural networks are deep learning models specifically designed to process data with temporal and spatial correlations, such as dynamic parameter changes in the battery. Multiphysics coupling modeling on cloud servers refers to a modeling method that considers the interaction of multiple physical phenomena within the battery, such as the thermo-electro-chemical coupling effect. This model can more accurately reflect the actual operating state of the battery. Quantum genetic algorithms are optimization algorithms that combine quantum computing and genetic algorithms. They search for optimal solutions by simulating biological genetic processes and the superposition characteristics of quantum states, and are used to optimize the energy transfer paths of intelligent bidirectional energy routing devices to achieve energy balance between battery cells. The pulse repair unit actively repairs SEI film damage by generating kHz-level microcurrents. The SEI film is a protective film inside the battery, and its damage affects battery performance and lifespan. The collaborative control module, through linkage with the BMS, PCS, and EMS, achieves system-level thermal management, energy routing, and strategy optimization. The BMS is responsible for the daily management of the battery, the PCS for power conversion, and the EMS for energy distribution and management.

[0025] Preferably, the construction process of the spatiotemporal convolutional neural network includes preprocessing the battery data, such as normalization and denoising, and then inputting it into the network for training. During training, a large amount of battery operating data is used as samples. By adjusting the network's weights and bias parameters, the network can accurately identify changes in battery state. In the data processing, a sliding window technique is used to segment the time-series data so that the network can learn the temporal correlation of the data. The energy transfer path optimization process of the intelligent bidirectional energy routing device includes initializing the population, i.e., randomly generating a set of possible paths, and then iteratively optimizing the paths through a quantum genetic algorithm, including selection, crossover, and mutation operations, until the optimal energy transfer path is found. In the process of formulating the repair strategy, the cloud server generates personalized repair strategies based on the battery's real-time state and historical data, combined with the simulation results of the digital twin platform. These strategies not only consider the battery's current state but also predict future state changes to achieve the best repair effect.

[0026] In some embodiments, the pulse repair unit triggers repair when the AI-predicted lithium plating exceeds a dynamic threshold or every 500 cycles, injecting an asymmetric microcurrent waveform to optimize the SEI film structure. The intelligent bidirectional energy routing device, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through a multi-modal balanced topology, suppressing parameter discretization between cells.

[0027] It should be noted that the pulse repair unit triggers repair when the AI ​​predicts that the lithium plating level exceeds the dynamic threshold or every 500 cycles. It injects an asymmetric microcurrent waveform to optimize the SEI film structure. This is to promptly repair the SEI film after the battery shows potential lithium plating risk or reaches a certain number of cycles, preventing further deterioration of battery performance. The intelligent bidirectional energy routing device, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through a multi-modal balanced topology, suppressing parameter discretization between cells. Its purpose is to reduce performance differences between cells by dynamically adjusting the energy distribution among them, thereby extending the overall lifespan of the battery pack.

[0028] Specifically, the asymmetric microcurrent waveform of the pulse repair unit is a special current signal whose waveform is asymmetrical in both the positive and negative half-cycles. This design helps to more effectively repair the SEI film. The SEI film is a solid electrolyte interface film inside the battery, which has a significant impact on battery performance and lifespan. When the AI ​​predicts that the lithium plating amount exceeds the dynamic threshold, it indicates that the lithium-ion deposition process inside the battery may have an anomaly. At this time, triggering the pulse repair unit to repair it can reduce the negative impact of lithium plating on battery performance. The multimodal balancing topology of the intelligent bidirectional energy routing device refers to the device's ability to use multiple modes for energy transfer and balancing based on different cell states. Cell internal resistance, temperature, and capacity differences are key factors affecting battery pack performance. By dynamically monitoring these parameters and adjusting the energy transfer strategy accordingly, the performance differences between cells can be effectively reduced, improving the overall performance and lifespan of the battery pack.

[0029] Preferably, the triggering conditions of the pulse repair unit can be adjusted according to the specific usage and performance requirements of the battery. For example, the dynamic threshold can be adaptively adjusted based on the battery type, usage environment, and historical performance data to ensure that repair is triggered at the optimal time. During the repair process, the specific parameters of the asymmetric microcurrent waveform, such as frequency, amplitude, and waveform shape, can also be optimized according to the current state of the battery. The energy transfer path of the intelligent bidirectional energy routing device can be optimized using a quantum genetic algorithm, which iteratively searches for the optimal energy transfer path by simulating a biological genetic process. In each iteration, the algorithm adjusts the energy transfer path based on changes in parameters such as the internal resistance, temperature, and capacity of the battery cells to achieve optimal energy balance between cells. Furthermore, the response time of the device can be optimized to the second level to quickly adapt to changes in the state of the battery cells and ensure the stable performance of the battery pack.

[0030] In some embodiments, the spatiotemporal convolutional neural network tracks the causal chain between material interface phase transitions and macroscopic performance degradation by decoupling the thermo-electro-chemical coupling effect. The digital twin platform synchronizes the evolution of microscopic parameters such as cell internal resistance and polarization characteristics with real-time monitoring data to generate precise repair instructions.

[0031] It should be noted that spatiotemporal convolutional neural networks, by decoupling the thermo-electro-chemical coupling effect, trace the causal chain between material interface phase transitions and macroscopic performance degradation. This means that the network can analyze the complex physicochemical processes inside the battery and identify the relationship between these processes and battery performance degradation. The digital twin platform synchronizes the evolution of microscopic parameters such as cell internal resistance and polarization characteristics with real-time monitoring data to generate precise repair instructions. This demonstrates that the digital twin platform can reflect the microscopic changes inside the battery in real time and formulate precise repair strategies accordingly to ensure stable battery performance and extend battery life.

[0032] Specifically, spatiotemporal convolutional neural networks (SCNNs) are deep learning models specifically designed to process data with temporal and spatial correlations. In battery management systems, they can analyze the thermo-electro-chemical coupling effects within the battery—the interactions between heat, current, and chemical reactions. These effects often lead to battery performance degradation, such as phase transitions at material interfaces (structural changes in the battery's internal materials) and macroscopic performance degradation, such as a decrease in battery capacity. A digital twin platform is a virtual model that can synchronize microscopic parameters within the battery in real time, such as cell resistance and polarization characteristics, reflecting the electrochemical reaction state within the battery. By combining these microscopic parameters with real-time monitoring data, the digital twin platform can generate precise repair instructions, guiding the intelligent repair module to perform targeted repair operations.

[0033] Preferably, the construction process of the spatiotemporal convolutional neural network includes steps such as data preprocessing, feature extraction, and model training. In the data preprocessing stage, the collected battery data needs to be normalized and denoised to improve data quality. In the feature extraction stage, the network automatically identifies key features related to battery performance degradation, such as current density, temperature gradient, and chemical reaction rate. In the model training stage, a large amount of historical and real-time data is used to train the network, enabling it to accurately predict changes in battery performance. The construction of the digital twin platform requires establishing physical and electrochemical models of the battery and fusing these models with real-time monitoring data. The platform dynamically adjusts repair instructions based on the model's prediction results and real-time data feedback to achieve precise control of the battery state. For example, when an abnormal change in the cell's internal resistance is detected, the platform generates corresponding repair instructions based on the trend and magnitude of the internal resistance change, guiding the intelligent repair module to perform micro-balancing or pulse repair operations to restore battery performance.

[0034] In some embodiments, the collaborative control module triggers the pre-start of the fire protection system when repairing an anomaly, and switches to a low-rate charging and discharging mode during the repair period; The system supports cross-brand plug-and-play functionality, is compatible with multi-cluster parallel configurations, and requires no modification to the existing energy storage architecture.

[0035] It should be noted that the collaborative control module triggers the pre-start of the fire suppression system when repairing anomalies and switches to a low-rate charge / discharge mode during the repair period. This is to ensure safety during battery repair and prevent potential dangerous situations such as thermal runaway. Simultaneously, the system supports cross-brand plug-and-play functionality and is compatible with multi-cluster parallel configurations without requiring modifications to existing energy storage architectures. This demonstrates the system's high versatility and flexibility, enabling easy integration into different energy storage systems without large-scale modifications to existing systems.

[0036] Specifically, the collaborative control module is a crucial component of the entire system, responsible for coordinating the operations between various modules to ensure the overall operational efficiency and safety of the system. Triggering pre-activation of the fire suppression system during anomaly repair means that when the system detects an abnormal battery condition that may cause safety issues, it will preemptively activate fire suppression systems, such as cooling devices or fire extinguishing equipment, to prevent accidents. Switching to a low-rate charge / discharge mode means that during battery repair, to reduce the burden on the battery and avoid further damage, the system automatically adjusts the charge / discharge rate to a rate lower than the normal operating rate. Cross-brand plug-and-play means that the system is compatible with batteries of different brands and models without complex adaptations and adjustments. Compatibility with multi-cluster parallel configurations means that the system can support the parallel connection of multiple battery clusters, which can improve the capacity and flexibility of the energy storage system. No need to modify existing energy storage architecture means that the system can be directly connected to existing energy storage systems as an external device without requiring large-scale modifications or upgrades to the existing system.

[0037] Preferably, in the collaborative control module, the pre-start of the fire suppression system can be achieved by setting a series of safety thresholds. When parameters such as battery temperature, voltage, or current exceed these thresholds, the system will automatically determine that there may be a safety risk and trigger the pre-start of the fire suppression system. Switching to low-rate charge / discharge modes can be achieved by adjusting the parameter settings of the battery management system (BMS), for example, reducing the charge / discharge rate from the normal 1C to 0.5C or lower to reduce the energy input and output of the battery during the repair process. For cross-brand plug-and-play functionality, the system can achieve compatibility with batteries from different brands through standardized interfaces and communication protocols. These interfaces and protocols can include battery connection methods, data transmission formats, and control signals. Regarding compatibility with multi-cluster parallel configurations, the system can achieve unified management and control of multiple battery clusters through an intelligent energy management system (EMS). The EMS can dynamically adjust the energy distribution according to the status and needs of each battery cluster to ensure the stable operation of the entire energy storage system. In addition, to further improve the system's flexibility and adaptability, the system can also provide a user-friendly configuration interface, allowing users to make personalized settings and adjustments according to their own needs and battery types.

[0038] In some embodiments, the pulse repair unit outputs a microcurrent amplitude of less than 0.1C, and the duration is 10%-30% of the battery resting period.

[0039] It should be noted that the battery repair method of this invention is a systematic repair process. It aims to collect multi-dimensional battery data in real time, utilize an edge computing terminal to analyze hidden failure factors and output safety warnings, and then use a cloud server to integrate historical data and a digital twin model to predict battery life degradation trajectories and formulate repair strategies. When the lithium plating exceeds a dynamic threshold or reaches a preset number of cycles, a pulse repair unit and an intelligent bidirectional energy routing device are activated, simultaneously implementing SEI film optimization and cell-level balancing. Based on a dynamic optimization strategy for repair effectiveness, the edge computing terminal and cloud server achieve autonomous algorithm iteration and system-level collaborative management. This method not only effectively repairs batteries but also improves repair efficiency and effectiveness through dynamic optimization strategies.

[0040] Specifically, real-time multi-dimensional battery data acquisition refers to collecting data such as current, voltage, temperature, gas, internal resistance, smoke, and pressure from the battery through a multi-dimensional sensing module. This data reflects the battery's operating status and forms the basis for subsequent diagnosis and repair. Edge computing terminal analysis of latent failure factors involves using deep reinforcement learning algorithms and spatiotemporal convolutional neural networks to analyze the collected data and identify potential factors that may lead to battery failure. Safety warnings are alerts issued based on these analysis results to remind users of potential battery risks. Cloud server fusion of historical data and digital twin models combines historical battery operating data with a physical model-based digital twin model to predict the battery's lifespan degradation trajectory. Repair strategies are specific repair measures formulated based on these predictions. The activation conditions for the pulse repair unit and intelligent bidirectional energy routing device are that the lithium deposition exceeds a dynamic threshold or reaches a preset number of cycles. This means that when the lithium-ion deposition inside the battery exceeds a certain limit or the battery has undergone a certain number of charge-discharge cycles, repair is required. SEI film optimization and cell-level balancing refer to repairing the SEI film through the pulse repair unit and balancing the energy of the cell through the intelligent bidirectional energy routing device. Dynamic optimization strategies refer to adjusting repair strategies based on repair results to achieve better outcomes. Autonomous algorithm iteration between edge computing terminals and cloud servers refers to optimizing algorithms based on repair results to improve accuracy and efficiency. System-level collaborative management refers to the collaborative management of the entire energy storage system during the repair process to ensure stable system operation.

[0041] Preferably, the real-time acquisition of multi-dimensional battery data can be achieved by setting the data acquisition frequency and accuracy. For example, current and voltage data can be acquired once per second, while temperature and internal resistance data can be acquired once per minute. When the edge computing terminal analyzes latent failure factors, different algorithm models can be set to identify different failure factors, such as overcharging, over-discharging, and overheating. Safety warnings can be triggered by setting thresholds; when data exceeds these thresholds, an alarm is issued. When the cloud server integrates historical data with the digital twin model, machine learning algorithms can be used to analyze historical data, and the repair strategy can be adjusted based on the prediction results of the digital twin model. The activation condition of the pulse repair unit can be determined by monitoring changes in lithium plating; when lithium plating exceeds a preset dynamic threshold, repair is automatically initiated. The energy transfer path of the intelligent bidirectional energy routing device can be optimized using quantum genetic algorithms to achieve the best energy balance effect. The evaluation of the repair effect can be achieved by monitoring battery performance parameters, such as capacity recovery rate and internal resistance reduction rate. Dynamic optimization strategies can be adjusted based on these evaluation results, such as adjusting the parameters of the pulse repair unit or optimizing the path of the energy routing device. The autonomous iteration of algorithms between edge computing terminals and cloud servers can be achieved through a feedback mechanism, optimizing the algorithms based on the repair results. System-level collaborative management can be achieved by setting different management strategies, such as adjusting the charging and discharging strategies of the energy storage system during repairs to reduce the impact on system operation.

[0042] The above embodiments of the present invention have the following beneficial effects: 1. By collecting battery current, voltage, temperature, gas, internal resistance, smoke, and pressure data through a multi-dimensional sensing module, and combining the deep reinforcement learning algorithm and spatiotemporal convolutional neural network integrated into the edge computing terminal, the battery dynamic parameters can be analyzed in real time and a lithium plating kinetic model can be established. This allows for the early detection of micro-damage such as lithium plating, effectively solving the problem of lagging traditional technical detection methods. It realizes the transformation from macro-parameter monitoring to early warning of micro-damage, greatly improving the safety and reliability of battery management, advancing the safety intervention window to the T1-T2 stage, and avoiding safety accidents such as thermal runaway.

[0043] 2. The intelligent repair module includes an intelligent bidirectional energy routing device optimized by a quantum genetic algorithm and a pulse repair unit. The former, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through multimodal equalization topology to suppress parameter discretization between cells. The latter can generate kHz-level microcurrents to actively repair SEI film damage. This active repair method overcomes the problems of low efficiency and inability to suppress parameter discretization between cells in traditional passive equalization technology. It achieves simultaneous cell-level micro-repair and system-level thermal-electric synergistic management, effectively extending battery life, reducing the total life cycle cost of the energy storage system, and improving the economics of battery management.

[0044] 3. The system adopts a cloud-edge collaborative architecture, integrating a multi-physics digital twin model. The cloud server generates battery life prediction and adaptive equalization repair strategies through multi-physics coupled modeling and deep reinforcement learning, and constructs a digital twin platform integrating electrochemical-physical simulation. This enables heterogeneous data collaborative processing between edge computing terminals and the cloud server, and links with BMS, PCS, and EMS for system-level thermal management, energy routing, and strategy optimization. This overcomes the bottlenecks of traditional technical model accuracy limitations and lack of collaborative management, improves the accuracy of battery life prediction, optimizes the management strategy of the entire energy storage system, reduces repair energy consumption, reduces the occurrence of secondary failures, and improves the overall performance and safety of the energy storage system.

[0045] In some embodiments, an AI-powered intelligent diagnostic and repair method for an external battery is provided, the method comprising: Real-time collection of multi-dimensional battery data, analysis of hidden failure factors through edge computing terminals, and output of safety warnings; The cloud server integrates historical data with digital twin models to predict lifespan degradation trajectories and formulate repair strategies; When the amount of lithium plating exceeds the dynamic threshold or reaches 500 cycles, the pulse repair unit and intelligent bidirectional energy routing device are activated to simultaneously implement SEI film optimization and cell-level balancing. Based on the dynamic optimization strategy of repair effect, the algorithm of edge computing terminal and cloud server can be autonomously iterated and system-level collaboratively managed.

[0046] It should be noted that the battery repair method of this invention is a systematic repair process. It aims to collect multi-dimensional battery data in real time, utilize an edge computing terminal to analyze hidden failure factors and output safety warnings, and then use a cloud server to integrate historical data and a digital twin model to predict battery life degradation trajectories and formulate repair strategies. When the lithium plating exceeds a dynamic threshold or reaches a preset number of cycles, a pulse repair unit and an intelligent bidirectional energy routing device are activated, simultaneously implementing SEI film optimization and cell-level balancing. Based on a dynamic optimization strategy for repair effectiveness, the edge computing terminal and cloud server achieve autonomous algorithm iteration and system-level collaborative management. This method not only effectively repairs batteries but also improves repair efficiency and effectiveness through dynamic optimization strategies.

[0047] Specifically, real-time multi-dimensional battery data acquisition refers to collecting data such as current, voltage, temperature, gas, internal resistance, smoke, and pressure from the battery through a multi-dimensional sensing module. This data reflects the battery's operating status and forms the basis for subsequent diagnosis and repair. Edge computing terminal analysis of latent failure factors involves using deep reinforcement learning algorithms and spatiotemporal convolutional neural networks to analyze the collected data and identify potential factors that may lead to battery failure. Safety warnings are alerts issued based on these analysis results to remind users of potential battery risks. Cloud server fusion of historical data and digital twin models combines historical battery operating data with a physical model-based digital twin model to predict the battery's lifespan degradation trajectory. Repair strategies are specific repair measures formulated based on these predictions. The activation conditions for the pulse repair unit and intelligent bidirectional energy routing device are that the lithium deposition exceeds a dynamic threshold or reaches a preset number of cycles. This means that when the lithium-ion deposition inside the battery exceeds a certain limit or the battery has undergone a certain number of charge-discharge cycles, repair is required. SEI film optimization and cell-level balancing refer to repairing the SEI film through the pulse repair unit and balancing the energy of the cell through the intelligent bidirectional energy routing device. Dynamic optimization strategies refer to adjusting repair strategies based on repair results to achieve better outcomes. Autonomous algorithm iteration between edge computing terminals and cloud servers refers to optimizing algorithms based on repair results to improve accuracy and efficiency. System-level collaborative management refers to the collaborative management of the entire energy storage system during the repair process to ensure stable system operation.

[0048] Preferably, the real-time acquisition of multi-dimensional battery data can be achieved by setting the data acquisition frequency and accuracy. For example, current and voltage data can be acquired once per second, while temperature and internal resistance data can be acquired once per minute. When the edge computing terminal analyzes latent failure factors, different algorithm models can be set to identify different failure factors, such as overcharging, over-discharging, and overheating. Safety warnings can be triggered by setting thresholds; when data exceeds these thresholds, an alarm is issued. When the cloud server integrates historical data with the digital twin model, machine learning algorithms can be used to analyze historical data, and the repair strategy can be adjusted based on the prediction results of the digital twin model. The activation condition of the pulse repair unit can be determined by monitoring changes in lithium plating; when lithium plating exceeds a preset dynamic threshold, repair is automatically initiated. The energy transfer path of the intelligent bidirectional energy routing device can be optimized using quantum genetic algorithms to achieve the best energy balance effect. The evaluation of the repair effect can be achieved by monitoring battery performance parameters, such as capacity recovery rate and internal resistance reduction rate. Dynamic optimization strategies can be adjusted based on these evaluation results, such as adjusting the parameters of the pulse repair unit or optimizing the path of the energy routing device. The autonomous iteration of algorithms between edge computing terminals and cloud servers can be achieved through a feedback mechanism, optimizing the algorithms based on the repair results. System-level collaborative management can be achieved by setting different management strategies, such as adjusting the charging and discharging strategies of the energy storage system during repairs to reduce the impact on system operation.

[0049] In some embodiments, during the safety intervention window period of the T1-T2 stage, cells with internal resistance dispersion > 5% or capacity decay rate > 3% are prioritized for repair. After the repair is completed, the recovery degree of cell polarization characteristics is verified by digital twin model. If the threshold is not reached, a second repair is triggered.

[0050] It should be noted that, within the safe intervention window of the T1-T2 stage, the battery repair method of this invention prioritizes repairing cells with an internal resistance dispersion greater than 5% or a capacity decay rate greater than 3%. This strategy aims to intervene at an early stage when battery performance begins to decline significantly, preventing further performance degradation. After repair, the recovery degree of cell polarization characteristics is verified using a digital twin model; if the threshold is not reached, a secondary repair is triggered. This step ensures the reliability and effectiveness of the repair effect; precise verification through the digital twin model ensures that the cell performance is restored to a satisfactory level.

[0051] Specifically, the T1-T2 stage refers to the period between when battery performance begins to show slight changes from its normal state and when these changes become irreversible. During this stage, the battery's internal resistance dispersion and capacity decay rate are two key performance indicators. Internal resistance dispersion refers to the degree of difference in internal resistance between individual cells in the battery pack. When this difference exceeds 5%, it indicates that the battery pack's balance is beginning to be compromised. Capacity decay rate refers to the rate at which the battery's capacity decreases relative to its initial capacity. When this rate exceeds 3%, it indicates that the battery's performance is beginning to decline significantly. A digital twin model is a virtual model that can reflect the battery's physical state in real time and verify the repair effect through simulation and analysis. Polarization characteristics refer to the potential changes on the electrode surface during charging and discharging, reflecting the battery's internal reaction state. Verifying the recovery of polarization characteristics through a digital twin model ensures that the battery's internal reaction state returns to normal after the repair operation.

[0052] Preferably, when implementing the repair strategy, a priority queue can be set up, marking cells with internal resistance dispersion greater than 5% or capacity decay rate greater than 3% as high priority. During the repair process, these high-priority cells are repaired first. After repair, the repaired cells are evaluated using a digital twin model. The evaluation process includes comparing the polarization characteristics of the repaired cells with thresholds set in the model. If the polarization characteristic recovery does not reach the set threshold, for example, if the recovery is less than 90%, a secondary repair is triggered. The secondary repair can use different parameter settings, such as adjusting the frequency or amplitude of the microcurrent waveform of the pulse repair unit, or optimizing the energy transfer path of the intelligent bidirectional energy routing device. Through these refined steps, it can be ensured that the performance of each cell is effectively restored, thereby improving the performance and lifespan of the entire battery pack.

[0053] In some embodiments, the intelligent bidirectional energy routing device dynamically optimizes the energy transfer path through a quantum genetic algorithm, adjusting the energy distribution between battery cells with a response time of seconds.

[0054] It should be noted that, within the safe intervention window of the T1-T2 stage, the battery repair method of this invention prioritizes repairing cells with an internal resistance dispersion greater than 5% or a capacity decay rate greater than 3%. This strategy aims to intervene at an early stage when battery performance begins to decline significantly, preventing further performance degradation. After repair, the recovery degree of cell polarization characteristics is verified using a digital twin model; if the threshold is not reached, a secondary repair is triggered. This step ensures the reliability and effectiveness of the repair effect; precise verification through the digital twin model ensures that the cell performance is restored to a satisfactory level.

[0055] Specifically, the T1-T2 stage refers to the period between when battery performance begins to show slight changes from its normal state and when these changes become irreversible. During this stage, the battery's internal resistance dispersion and capacity decay rate are two key performance indicators. Internal resistance dispersion refers to the degree of difference in internal resistance between individual cells in the battery pack. When this difference exceeds 5%, it indicates that the battery pack's balance is beginning to be compromised. Capacity decay rate refers to the rate at which the battery's capacity decreases relative to its initial capacity. When this rate exceeds 3%, it indicates that the battery's performance is beginning to decline significantly. A digital twin model is a virtual model that can reflect the battery's physical state in real time and verify the repair effect through simulation and analysis. Polarization characteristics refer to the potential changes on the electrode surface during charging and discharging, reflecting the battery's internal reaction state. Verifying the recovery of polarization characteristics through a digital twin model ensures that the battery's internal reaction state returns to normal after the repair operation.

[0056] Preferably, when implementing the repair strategy, a priority queue can be set up, marking cells with internal resistance dispersion greater than 5% or capacity decay rate greater than 3% as high priority. During the repair process, these high-priority cells are repaired first. After repair, the repaired cells are evaluated using a digital twin model. The evaluation process includes comparing the polarization characteristics of the repaired cells with thresholds set in the model. If the polarization characteristic recovery does not reach the set threshold, for example, if the recovery is less than 90%, a secondary repair is triggered. The secondary repair can use different parameter settings, such as adjusting the frequency or amplitude of the microcurrent waveform of the pulse repair unit, or optimizing the energy transfer path of the intelligent bidirectional energy routing device. Through these refined steps, it can be ensured that the performance of each cell is effectively restored, thereby improving the performance and lifespan of the entire battery pack.

[0057] In some embodiments, the system-level collaborative management includes dynamic adjustment of liquid cooling flow, cross-module energy routing compensation, and optimization of EMS charging and discharging strategies. All repair operations are conducted on the premise of not affecting the grid-connected operation of the energy storage system, and the energy consumption of repair is less than 0.5% of the total system capacity.

[0058] It should be noted that the battery repair method of this invention includes system-level collaborative management, which includes dynamic adjustment of liquid cooling flow, cross-module energy routing compensation, and optimization of the energy management system (EMS) charging and discharging strategy. These measures aim to ensure the stable operation and efficient management of the entire energy storage system during the repair operation. All repair operations are conducted without affecting the grid-connected operation of the energy storage system, and the repair energy consumption accounts for less than 0.5% of the total system capacity. This requirement ensures the efficiency and economy of the repair process while reducing interference with the normal operation of the system.

[0059] Specifically, dynamic liquid cooling flow rate adjustment refers to dynamically adjusting the coolant flow rate of the liquid cooling system based on the real-time temperature and load of the battery pack to maintain the battery pack within its optimal operating temperature range. Cross-module energy routing compensation refers to the use of intelligent energy routing devices in multi-module energy storage systems to transfer energy between different modules, balancing energy differences between modules and improving overall system efficiency. Energy Management System (EMS) charge / discharge strategy optimization refers to dynamically adjusting the charge / discharge strategy based on the real-time status of the battery pack and system requirements to optimize battery life and system performance. The implementation parameters for these measures include real-time data such as battery pack temperature, voltage, and current, as well as system load requirements and operating status. Through monitoring and analysis of these parameters, the system can automatically adjust the liquid cooling flow rate, energy transfer path, and charge / discharge strategy to achieve optimal system performance.

[0060] Preferably, in the dynamic adjustment of liquid cooling flow rate, a temperature sensor can be set to monitor the real-time temperature of the battery pack and dynamically adjust the coolant flow rate according to a preset temperature threshold. For example, when the battery pack temperature exceeds 40 degrees Celsius, the coolant flow rate is increased; when the temperature is below 35 degrees Celsius, the coolant flow rate is decreased. In cross-module energy routing compensation, an intelligent bidirectional energy routing device monitors the voltage and current of each module, calculates the energy difference between modules, and dynamically adjusts the energy transfer path to achieve energy balance between modules. In the optimization of charging and discharging strategies in the energy management system (EMS), the charging and discharging current and voltage can be dynamically adjusted according to the battery pack's state of charge (SOC) and system load requirements. For example, when the battery pack's SOC is below 20%, the charging current is increased; when the system load requirement is high, the discharging strategy is optimized to improve system efficiency. These optimization measures can be implemented through preset algorithms and models to ensure stable operation and efficient management of the system during repair operations.

[0061] In some embodiments, the energy transfer path optimized by the quantum genetic algorithm reduces the inter-cell capacity standard deviation to below 1.23%, voltage dispersion to more than 45%, internal resistance dispersion to more than 38%, and temperature dispersion to more than 51%.

[0062] It should be noted that the energy transfer path optimized by the quantum genetic algorithm in this invention can significantly reduce the dispersion of capacity, voltage, internal resistance, and temperature among battery cells. This achievement demonstrates that by optimizing the energy transfer path, the performance differences between battery cells can be effectively reduced, thereby extending the overall lifespan of the battery pack and improving its performance stability. Specifically, the algorithm reduces the standard deviation of capacity among battery cells to below 1.23%, voltage dispersion to over 45%, internal resistance dispersion to over 38%, and temperature dispersion to over 51%. These data reflect the algorithm's high efficiency in balancing battery pack performance.

[0063] Specifically, the quantum genetic algorithm is an advanced optimization algorithm that combines the efficiency of quantum computing with the global search capability of genetic algorithms. In battery management systems, this algorithm is used to optimize the energy transfer paths of an intelligent bidirectional energy routing device. An energy transfer path refers to the path along which energy is transferred from one cell to another within a battery pack. By optimizing these paths, energy balance between cells can be achieved, reducing the overall performance degradation of the battery pack caused by performance differences between cells. The standard deviation of cell capacity refers to the degree of difference in capacity among the individual cells in the battery pack; a lower standard deviation indicates more balanced capacity among the cells. Voltage, internal resistance, and temperature dispersion refer to the degree of difference in voltage, internal resistance, and temperature between cells, respectively; lower dispersion indicates greater consistency of these parameters among the cells, contributing to improved battery pack performance and lifespan.

[0064] Preferably, when implementing the quantum genetic algorithm, the population first needs to be initialized, i.e., a set of possible energy transfer paths is randomly generated. Then, by evaluating the fitness of each path, the path with high fitness is selected for crossover and mutation operations to generate new paths. The fitness function can be designed based on the differences in capacity, voltage, internal resistance, and temperature between cells; for example, fitness can be inversely proportional to the dispersion of parameters between cells. In the crossover operation, parts of two paths are exchanged to generate new path combinations. In the mutation operation, certain parts of the paths are randomly changed to explore new solution spaces. Through multiple iterations, the algorithm gradually converges to the optimal energy transfer path. In practical applications, the algorithm's input parameters include real-time capacity, voltage, internal resistance, and temperature data of the cells, which are collected through a multi-dimensional sensing module. The algorithm's output is the optimized energy transfer path, which is used to control an intelligent bidirectional energy routing device to achieve energy balance between cells. For example, when a cell's voltage is detected to be lower than other cells, the algorithm calculates the optimal path to transfer energy from the high-voltage cell to the low-voltage cell, thereby reducing the voltage difference between cells.

[0065] like Figure 2 As shown, when the amount of lithium plating exceeds the dynamic threshold or reaches the preset number of cycles, the pulse repair unit and energy routing device are activated to simultaneously implement SEI film optimization and cell-level balancing. The comparison before and after repair shows that lithium plating is reduced by 76.4%. The intelligent bidirectional energy routing device, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through a multi-modal balanced topology, suppressing parameter discretization between cells.

[0066] like Figure 3As shown, the balancing strategy dynamically optimizes the energy transfer path through a quantum genetic algorithm, adjusting the energy distribution between cells with a response time of seconds. The comparison of the effects before and after balancing shows that the standard deviation of the cell capacity before balancing decreased from 2.63% (range 10.0%) to 1.23% (range 4.6%), and the dispersion of voltage, internal resistance, and temperature decreased by 45%, 38%, and 51%, respectively.

[0067] The spatiotemporal convolutional neural network decouples the thermo-electro-chemical coupling effect, tracks the causal chain between material interface phase transitions and macroscopic performance degradation, collects multi-dimensional battery data in real time, and analyzes and issues warnings about hidden failure factors through edge computing terminals. like Figure 4 As shown in Table 1, thermal runaway predictions are made and risk level assessments and countermeasures are given during the safety intervention window period of T1-T2. Table 1. Prediction and Early Warning Table for Thermal Runaway Classification

[0068] The cloud server integrates historical data with a digital twin model to predict battery life degradation trajectories and formulate repair strategies. The digital twin platform synchronizes the evolution of microscopic parameters such as cell internal resistance and polarization characteristics with real-time monitoring data to generate precise repair instructions.

[0069] Based on the dynamic optimization strategy of repair effect, the edge-cloud algorithm can be autonomously iterated and the system-level collaborative management can be realized.

[0070] The system-level collaborative management includes dynamic adjustment of liquid cooling flow, cross-module energy routing compensation, and EMS charging and discharging strategy optimization. Specifically, the pulse repair uses a microcurrent waveform with a frequency of 1-10 kHz and an amplitude of less than 0.1C, and the duration is 10-30% of the battery's resting period; Specifically, the collaborative control module triggers measures such as pre-start of the fire protection system when an anomaly is predicted, and switches to a low-rate charging and discharging mode during the repair period; After the repair is completed, the recovery degree of cell polarization characteristics is verified by digital twin model. If the threshold is not reached, a second repair is triggered.

[0071] All repair operations are conducted on the premise of not affecting the grid-connected operation of the energy storage system, and the energy consumption of repair is less than 0.5% of the total system capacity.

[0072] The system supports plug-and-play across brands and is compatible with multi-cluster parallel configurations without requiring modifications to existing energy storage architectures. Furthermore, other energy storage systems can be configured with any number of battery cells.

[0073] To verify the accuracy and effectiveness of the electrochemical energy storage outlier lithium iron phosphate cell diagnostic algorithm of this invention, we adopted a specific experimental setup. In the experiment, we selected the iESS-CAB-W215H smart energy storage air-cooled integrated cabinet provided by Hongzheng Energy Storage. This system consists of 240 lithium iron phosphate batteries with a rated capacity of 280Ah manufactured by Ruipu Lanjun Energy Co., Ltd.

[0074] Furthermore, the storage medium in the embodiments of this application stores program instructions capable of implementing all the above methods. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0075] The above description is merely an explanation of some preferred embodiments of the present invention and the technical principles employed. Those skilled in the art should understand that the scope of the invention as described in the embodiments of the present invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. An external battery AI intelligent diagnostic and repair system, characterized in that, include: The multi-dimensional sensing module is configured to collect data on the battery's current, voltage, temperature, gas, internal resistance, smoke, and pressure. The edge computing terminal integrates deep reinforcement learning algorithms and spatiotemporal convolutional neural networks to analyze battery dynamic parameters in real time, establish lithium plating kinetic models, and output latent degradation features and safety warnings. The cloud server is configured to generate battery lifetime prediction and adaptive equilibrium repair strategies through multi-physics coupled modeling and deep reinforcement learning, and to build a digital twin platform that integrates electrochemical-physical simulation. The intelligent repair module includes an intelligent bidirectional energy routing device optimized by a quantum genetic algorithm and a pulse repair unit; wherein the intelligent bidirectional energy routing device implements micro-operation balancing, and the pulse repair unit generates a microcurrent with a frequency of 1-10kHz to actively repair SEI film damage. The collaborative control module enables collaborative processing of heterogeneous data between edge computing terminals and cloud servers, and links the battery management system (BMS), power conversion system (PCS), and energy management system (EMS) for system-level thermal management, energy routing, and strategy optimization.

2. The system according to claim 1, characterized in that: The pulse repair unit is triggered to repair when the AI ​​predicts that the amount of lithium plating exceeds the dynamic threshold or every 500 cycles, injecting an asymmetric microcurrent waveform to optimize the SEI film structure. The intelligent bidirectional energy routing device, based on differences in cell internal resistance, temperature, and capacity, dynamically transfers energy through a multi-modal balanced topology, suppressing parameter discretization between cells.

3. The system according to claim 1, characterized in that: The spatiotemporal convolutional neural network traces the causal chain between material interface phase transitions and macroscopic performance degradation by decoupling the thermo-electrochemical coupling effect. The digital twin platform synchronizes the evolution of microscopic parameters such as cell internal resistance and polarization characteristics with real-time monitoring data to generate precise repair instructions.

4. The system according to claim 1, characterized in that: The collaborative control module triggers the pre-start of the fire protection system when repairing an anomaly, and switches to a low-rate charging and discharging mode during the repair period; The system supports cross-brand plug-and-play functionality, is compatible with multi-cluster parallel configurations, and requires no modification to the existing energy storage architecture.

5. The system according to claim 1, characterized in that: The pulse repair unit outputs a microcurrent with an amplitude of less than 0.1C, and the duration is 10%-30% of the battery's resting period.

6. A battery repair method based on the system according to any one of claims 1-5, characterized in that, Includes the following steps: Real-time collection of multi-dimensional battery data, analysis of hidden failure factors through edge computing terminals, and output of safety warnings; The cloud server integrates historical data with digital twin models to predict lifespan degradation trajectories and formulate repair strategies; When the amount of lithium plating exceeds the dynamic threshold or reaches 500 cycles, the pulse repair unit and intelligent bidirectional energy routing device are activated to simultaneously implement SEI film optimization and cell-level balancing. Based on the dynamic optimization strategy of repair effect, the algorithm of edge computing terminal and cloud server can be autonomously iterated and system-level collaboratively managed.

7. The method according to claim 6, characterized in that: During the safety intervention window of the T1-T2 phase, priority should be given to repairing cells with internal resistance dispersion >5% or capacity decay rate >3%. After the repair is completed, the recovery degree of cell polarization characteristics is verified by digital twin model. If the threshold is not reached, a second repair is triggered.

8. The method according to claim 6, characterized in that: The intelligent bidirectional energy routing device dynamically optimizes the energy transfer path through a quantum genetic algorithm, adjusting the energy distribution between battery cells with a response time of up to seconds.

9. The method according to claim 6, characterized in that: The system-level collaborative management includes dynamic adjustment of liquid cooling flow, cross-module energy routing compensation, and optimization of EMS charging and discharging strategies. All repair operations are conducted on the premise of not affecting the grid-connected operation of the energy storage system, and the energy consumption of repair is less than 0.5% of the total system capacity.

10. The method according to claim 8, characterized in that: The energy transfer path optimized by the quantum genetic algorithm reduces the standard deviation of capacity between cells to below 1.23%, voltage dispersion to more than 45%, internal resistance dispersion to more than 38%, and temperature dispersion to more than 51%.

Images