High and cold high-altitude battery charging method based on multi-mode perception and reinforcement learning

Abstract

The invention relates to the technical field of reinforcement learning, and discloses a high-cold high-altitude battery charging method based on multi-mode perception and reinforcement learning, which comprises the following steps: analyzing environmental characteristics of a high-cold high-altitude battery to determine a sensor type selection standard and a hardware adaptation system of the high-cold high-altitude battery, constructing a multi-mode sensing network of the high-cold and high-altitude battery; analyzing a battery-environment coupling relationship between the high-cold and high-altitude battery and the environment stress corresponding to the environment characteristics, and performing characteristic fusion on the multi-modal data; the battery charging state of the high-cold and high-altitude battery is analyzed; constructing a reinforcement learning framework of the high-cold high-altitude battery, constructing a target excitation function of the high-cold high-altitude battery, defining a control action set of the high-cold high-altitude battery, and integrating a charging strategy decision network; and outputting a target charging instruction of the high-cold high-altitude battery, and executing dynamic charging of the high-cold high-altitude battery. According to the invention, the charging safety and efficiency of the high-cold and high-altitude battery can be improved.

Description

Technical Field

[0001] This invention relates to a method for charging batteries in high-altitude and cold regions based on multimodal perception and reinforcement learning, belonging to the field of reinforcement learning technology. Background Technology

[0002] High-altitude and cold-weather battery charging refers to the specialized technologies and methods used to safely, efficiently, and extend the lifespan of various batteries (especially lithium-ion batteries) under the coupled conditions of extreme environments: frigid conditions (typically below -20°C, even -40°C) and high altitude conditions (typically above 3000 meters, characterized by thin air, low air pressure, strong ultraviolet radiation, and large diurnal temperature variations). This technology can better store these intermittent renewable energy sources, providing stable and clean electricity to base stations, weather stations, and villages in remote areas, reducing reliance on diesel generators.

[0003] Traditional high-altitude and cold-weather battery charging methods employ a constant-current, constant-voltage charging strategy. This involves rapid charging with a constant current, then switching to constant-voltage charging once the battery voltage reaches a set value, with the current gradually decreasing until fully charged. However, this method has significant limitations in extreme environments. For example, the constant-current, constant-voltage charging method proposed in patent CN115566294B experiences a charging efficiency drop of over 30% in environments below -30°C and carries the risk of lithium plating. The literature *Journal of PowerSources, 2020, 455: 227966* points out that the traditional method significantly increases battery internal resistance in low-pressure environments, leading to prolonged charging time and an increased probability of thermal runaway. This charging method is ill-suited to extreme environments, easily causing overcharging, undercharging, or even thermal runaway, thus affecting the charging efficiency and safety of high-altitude and cold-weather batteries. In contrast, this invention uses a multimodal sensor network to perceive the battery and environmental state in real time, constructs a battery-environment coupling relationship model, and uses reinforcement learning algorithms (such as DQN and DDPG) to make dynamic charging decisions, fundamentally solving the problems of poor adaptability and low safety of traditional methods in extreme environments. Summary of the Invention

[0004] This invention provides a method for charging batteries in high-altitude and cold regions based on multimodal perception and reinforcement learning, with the main purpose of improving the safety and efficiency of battery charging in such conditions.

[0005] To achieve the above objectives, this invention provides a high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning, comprising: The environmental characteristics of the high-altitude and cold-weather battery are analyzed to determine the sensor selection criteria and hardware adaptation system for the battery. Based on the sensor selection criteria and hardware adaptation system, a multimodal sensing network for the high-altitude and cold-weather battery is constructed. Multimodal data of the high-altitude and cold-weather battery are collected through the multimodal sensing network to analyze the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress corresponding to the environmental characteristics. Based on the battery-environment coupling relationship, feature fusion is performed on the multimodal data to obtain fused data. Based on the fused data, the ionization degree reduction coefficient, lithium ion mobility, and internal resistance change rate of the high-altitude and cold-weather battery are calculated to analyze the battery charging state of the high-altitude and cold-weather battery. The charging objective of the high-altitude and cold-weather battery is defined to construct the target excitation function of the high-altitude and cold-weather battery. The control action set of the high-altitude and cold-weather battery is defined, and the control action set and the target excitation function are integrated into a preset reinforcement learning framework to obtain the charging strategy decision network. Based on the battery charging state, the target charging command for the high-altitude and cold-weather battery is output through the charging strategy decision network, and dynamic charging of the high-altitude and cold-weather battery is executed based on the target charging command.

[0006] Optionally, the step of calculating the ionization degree reduction coefficient, lithium-ion mobility, and internal resistance change rate of the high-altitude and cold-weather battery based on the fused data includes: Extract the temperature sequence, air pressure sequence, electrolyte concentration sequence, electrolyte viscosity, and electrochemical characteristics from the fused data; The real-time degree of ionization of the high-altitude, cold-weather battery is calculated based on the temperature sequence, the air pressure sequence, and the electrolyte concentration sequence. Based on the real-time degree of ionization, the degree of ionization reduction coefficient of the high-altitude and cold-weather battery is determined; Determine the lithium-ion diffusion coefficient of the high-altitude, cold-weather battery; The lithium-ion mobility of the high-altitude and cold-weather battery is calculated based on the electrolyte viscosity and the lithium-ion diffusion system. Based on the electrochemical characteristics, the real-time internal resistance of the high-altitude and cold-weather battery was calculated. The rate of change of internal resistance of the high-altitude and cold-weather battery is determined based on the real-time internal resistance.

[0007] Optionally, the analysis of the state of charge of the high-altitude, cold-weather battery includes: Based on the lithium-ion mobility and internal resistance change rate of the high-altitude and cold-weather battery, the chargeability index of the high-altitude and cold-weather battery is calculated. The temperature compensation factor of the high-altitude and cold-weather battery is determined based on the ionization degree reduction coefficient of the battery. Based on the temperature compensation factor and the lithium-ion mobility, the lithium plating risk coefficient of the high-altitude and cold-weather battery is calculated. Calculate the temperature rise rate of the high-altitude, cold-weather battery to analyze the thermal runaway risk coefficient of the high-altitude, cold-weather battery; The total risk coefficient of the high-altitude and cold-weather battery is determined based on the lithium plating risk coefficient and the thermal runaway risk coefficient. The state of charge of the high-altitude and cold-weather battery is determined based on the charging acceptability index and the total risk coefficient.

[0008] Optionally, the analysis of the battery-environment coupling relationship between the high-altitude, cold-climate battery and the environmental stress corresponding to the environmental characteristics includes: The multimodal data corresponding to the high-altitude and cold-weather battery are preprocessed to obtain preprocessed multimodal data. The preprocessed multimodal data is discretized to obtain discretized multimodal data; Extract the discrete multimodal features from the discrete multimodal data; Calculate the correlation coefficient between the discretized multimodal characteristics and the environmental stress; Based on the correlation coefficient, a battery-environment dynamic coupling equation is constructed for the high-altitude and cold-weather battery and the environmental stress. Based on the battery-environment dynamic coupling equation, the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress is determined.

[0009] Optionally, constructing the battery-environment dynamic coupling equation for the high-altitude, cold-climate battery and the environmental stress based on the correlation coefficient includes: Based on the correlation coefficient, a linear influence matrix of the environmental stress on the high-altitude and cold-weather battery is constructed. Define the comprehensive impact index of the environmental stress on the high-altitude and cold-weather battery; Based on the comprehensive impact index, the nonlinear coupling interaction matrix of the environmental stress on the high-altitude and cold-weather battery is analyzed. Based on the discretized multimodal characteristics corresponding to the high-altitude and cold-climate battery, the self-evolution coefficient matrix of the high-altitude and cold-climate battery is determined. Based on the linear influence matrix, the nonlinear coupling interaction matrix, and the self-evolution coefficient matrix, a battery-environment dynamic coupling equation for the high-altitude and cold-climate battery and the environmental stress is constructed.

[0010] Optionally, the analysis of the environmental characteristics of batteries in high-altitude and cold regions includes: Clearly define the application scenarios for the high-altitude and cold-weather batteries; Define the geographical and climatic conditions that define the application scenario; Based on the defined geographical and climatic conditions, the environmental factors of the high-altitude and cold-climate battery are determined. Based on the aforementioned environmental factors, the environmental stress of the high-altitude and cold-weather battery is determined, wherein the environmental stress includes: primary environmental stress and secondary environmental stress; Based on the environmental stress, the environmental characteristics of the high-altitude and cold-weather battery are determined.

[0011] Optionally, determining the sensor selection criteria and hardware adaptation system for the high-altitude, cold-weather battery includes: Based on the environmental characteristics corresponding to the high-altitude and cold-weather battery, the operating temperature range and operating air pressure range of the sensor corresponding to the high-altitude and cold-weather battery are determined. Based on the operating temperature range, the low-temperature start-up performance, low-temperature accuracy curve, and low-temperature drift coefficient of the sensor are determined. Based on the low-temperature start-up performance, the low-temperature accuracy curve, and the low-temperature drift coefficient, the low-temperature selection criteria for the sensor are determined. Based on the operating air pressure range, the pressure resistance, electrical clearance, and creepage distance of the sensor are determined. Based on the pressure resistance, the electrical clearance, and the creepage distance, the low-pressure selection criteria for the sensor are determined. Based on the low-temperature selection criteria and the air pressure selection criteria, the sensor selection criteria for the high-altitude and cold-weather battery are determined. The hardware materials of the sensor are determined according to the sensor selection criteria. Based on the aforementioned hardware materials, the sensor's protective structure, electrical connection structure, signal conditioning structure, and thermal management mechanism are constructed. Based on the protective structure, the electrical connection structure, the signal conditioning structure, and the thermal management mechanism, a hardware adaptation system for the sensor is constructed.

[0012] Optionally, the construction of the multimodal sensing network for the high-altitude, cold-weather battery based on the sensor selection criteria and the hardware adaptation system includes: Based on the sensor selection criteria and the hardware adaptation system, the multimodal sensors and sensor network accessories for the high-altitude and cold-weather battery are determined. Construct a network layout diagram of the sensor network accessories and the multimodal sensor; Determine the connection method and communication protocol between the sensor network accessory and the multimodal sensor; Based on the network layout diagram, the connection method, and the communication protocol, a multimodal sensing network for the high-altitude and cold-weather battery is constructed.

[0013] Optionally, integrating the set of control actions and the target activation function into a preset reinforcement learning framework to obtain a charging policy decision network includes: Obtain historical battery data for the high-altitude, cold-weather battery corresponding to the target excitation function; Based on the historical battery data, the electrochemical and thermodynamic characteristics of the high-altitude and cold-weather battery were analyzed. Based on the electrochemical and thermodynamic characteristics, a battery model for the high-altitude and cold-weather battery is constructed. Based on the environmental characteristics of the high-altitude and cold-weather battery, a high-altitude and cold-weather environmental model of the battery is constructed. The battery model and the high-altitude and cold environment model are integrated to obtain a battery environment simulation model. Based on the target activation function, a reinforcement learning algorithm for the high-altitude and cold-weather battery is determined. Based on the reinforcement learning algorithm and the set of control actions, the learning network structure of the high-altitude and cold-weather battery is constructed. By combining the battery environment simulation model, the learning network structure, and the reinforcement learning framework, a charging strategy decision network for the high-altitude and cold-weather battery is integrated.

[0014] Optionally, the target excitation function for constructing the high-altitude, cold-weather battery includes: The charging target corresponding to the high-altitude and cold-weather battery is decomposed into multiple sub-targets; Construct multiple incentive functions for the multiple sub-objectives, wherein the multiple incentive functions include: a safety reward function, an efficiency reward function, and a lifetime reward function; Determine the importance weights of the aforementioned multiple activation functions; Based on the importance weights, the multiple activation functions are weighted and summed to obtain the target activation function.

[0015] Compared to the problems described in the background technology, this invention provides a scientific basis for sensor selection and hardware adaptation through in-depth analysis of the characteristics of high-altitude and cold environments. Complex environmental stresses such as low temperature, low air pressure, and strong radiation directly affect the electrochemical reaction kinetics and material stability of batteries. Through systematic research on these environmental characteristics, this invention can accurately identify key parameters affecting battery performance, such as temperature gradient, air pressure changes, and radiation intensity. This not only guides the selection of high-precision, high-reliability sensors, ensuring the accuracy and real-time nature of data acquisition, but also promotes the specialized design of hardware systems, such as low-temperature start-up circuits and air pressure compensation structures, fundamentally improving the adaptability and robustness of the BMS (Battery Management System) in extreme environments. Secondly, based on multimodal sensor networks and feature fusion technology, this invention achieves deep perception and precise characterization of battery status. By constructing a multimodal sensor network, this invention integrates multidimensional data such as electrical, thermal, mechanical, and gaseous data, and performs feature-level fusion based on a battery-environment coupling relationship model, effectively eliminating the limitations of a single sensor. This invention further calculates key internal parameters such as the ionization degree decrease coefficient, lithium-ion mobility, and internal resistance change rate, elevating the battery charging state to a comprehensive state vector encompassing electrochemical characteristics, thermodynamic behavior, and safety boundaries. This deep perception capability provides a high-quality, high-information-density data foundation for subsequent intelligent decision-making. Finally, the introduction of a reinforcement learning decision network evolves the charging strategy from fixed rules to dynamic intelligence, achieving optimal control under multi-objective optimization. This invention constructs charging objectives (such as speed, safety, and lifespan) as objective excitation functions and defines parameters such as charging current and voltage as a set of control actions. By using the fused battery state as input, the reinforcement learning network can evaluate the long-term rewards of different charging actions in real time and output the target charging command under the current operating condition. This dynamic charging strategy can intelligently balance fast charging and safety protection based on the actual battery state and environmental changes, effectively avoiding risks such as low-temperature lithium plating and high-voltage overcharging, significantly shortening charging time, and maximizing battery cycle life. Therefore, the high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning provided in this embodiment of the invention can improve the safety and efficiency of high-altitude and cold-weather battery charging. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a module for implementing the high-altitude, cold-weather battery charging system based on multimodal perception and reinforcement learning, according to an embodiment of the present invention.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning. The execution entity of this high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning according to an embodiment of the present invention. In this embodiment, the high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning includes: S1. Analyze the environmental characteristics of the high-altitude and cold-weather battery to determine the sensor selection criteria and hardware adaptation system for the high-altitude and cold-weather battery. Based on the sensor selection criteria and the hardware adaptation system, construct the multimodal sensing network of the high-altitude and cold-weather battery.

[0021] This invention, through analyzing the environmental characteristics of batteries operating at high altitudes and in cold conditions, can accurately identify key parameters affecting battery performance, such as temperature gradients, air pressure changes, and radiation intensity. This not only guides the selection of high-precision, high-reliability sensors, ensuring the accuracy and real-time nature of data acquisition, but also promotes the specialized design of hardware systems, such as low-temperature start-up circuits and air pressure compensation structures. The environmental characteristics refer to a set of external physical and chemical parameters that directly affect the battery's electrochemical reactions, physical structure, thermodynamic behavior, and safety boundaries, such as thermodynamic characteristics (ambient temperature, solar radiation, heat flux, etc.) and atmospheric physical characteristics (such as air pressure, air density, etc.).

[0022] As an embodiment of the present invention, the analysis of the environmental characteristics of batteries in high-altitude and cold regions includes: Clearly define the application scenarios for the high-altitude and cold-weather batteries; Define the geographical and climatic conditions that define the application scenario; Based on the defined geographical and climatic conditions, the environmental factors of the high-altitude and cold-climate battery are determined. Based on the aforementioned environmental factors, the environmental stress of the high-altitude and cold-weather battery is determined, wherein the environmental stress includes: primary environmental stress and secondary environmental stress; Based on the environmental stress, the environmental characteristics of the high-altitude and cold-weather battery are determined.

[0023] The application scenarios refer to the specific equipment, systems, and tasks served by high-altitude and cold-climate batteries, such as electric vehicles in plateau regions and high-altitude meteorological stations. The defined geographical area refers to the specific geographical region delineated based on the application scenario. The climatic conditions refer to the statistical laws governing the atmospheric state and its changing characteristics within the defined geographical area, summarized from long-term meteorological observation data, such as temperature and air pressure. The environmental factors refer to independent, measurable physical and chemical parameters extracted from climatic conditions and geographical features that can directly and indirectly affect battery performance, such as thermodynamic factors, atmospheric physical factors, and mechanical stress factors. The environmental stress refers to the disturbances, loads, and destructive effects of environmental factors on the battery's materials, structure, and electrochemical performance. The primary environmental stress refers to environmental stresses that have a decisive, global, and continuous impact on the battery's performance, lifespan, and safety, such as low-temperature stress and low atmospheric pressure. The secondary environmental stresses refer to environmental stresses that affect the battery, but whose impact is usually less than that of the primary stresses and only become significant under specific conditions, such as strong ultraviolet radiation stress and wind load stress.

[0024] Optionally, the defined geography can be determined using geographic information systems and remote sensing technology, and the climate conditions can be determined using global / regional climate models.

[0025] Optionally, the environmental stress of the high-altitude and cold-climate battery can be determined by stress mapping methods of physical-chemical models, such as thermo-electric coupling models, electrochemical-thermal coupling models, etc.

[0026] This invention, through defining the sensor selection criteria and hardware adaptation system for the high-altitude, cold-weather battery, ensures that the sensors and hardware platform themselves remain undamaged and function properly in environments with temperatures as low as -40°C or even lower, high ultraviolet radiation, strong pressure changes, and severe vibrations. The hardware adaptation system (such as heating and insulation modules, sealing structures, and vibration damping designs) further creates a locally comfortable microenvironment for the sensors, enabling stable operation. The sensor selection criteria refer to a set of systematic technical specifications and quantitative indicators developed to meet the performance, reliability, and safety requirements of high-altitude, cold-weather battery systems in specific application scenarios, used for evaluating and selecting various sensors. The hardware adaptation system refers to a comprehensive hardware solution designed around the selected sensors to ensure stable, reliable, and efficient operation in the specific physical, electrical, and thermodynamic environments of the high-altitude, cold-weather battery system, encompassing physical installation, electrical connections, signal conditioning, environmental protection, and thermal management.

[0027] As an embodiment of the present invention, determining the sensor selection criteria and hardware adaptation system for the high-altitude and cold-weather battery includes: Based on the environmental characteristics corresponding to the high-altitude and cold-weather battery, the operating temperature range and operating air pressure range of the sensor corresponding to the high-altitude and cold-weather battery are determined. Based on the operating temperature range, the low-temperature start-up performance, low-temperature accuracy curve, and low-temperature drift coefficient of the sensor are determined. Based on the low-temperature start-up performance, the low-temperature accuracy curve, and the low-temperature drift coefficient, the low-temperature selection criteria for the sensor are determined. Based on the operating air pressure range, the pressure resistance, electrical clearance, and creepage distance of the sensor are determined. Based on the pressure resistance, the electrical clearance, and the creepage distance, the low-pressure selection criteria for the sensor are determined. Based on the low-temperature selection criteria and the air pressure selection criteria, the sensor selection criteria for the high-altitude and cold-weather battery are determined. The hardware materials of the sensor are determined according to the sensor selection criteria. Based on the aforementioned hardware materials, the sensor's protective structure, electrical connection structure, signal conditioning structure, and thermal management mechanism are constructed. Based on the protective structure, the electrical connection structure, the signal conditioning structure, and the thermal management mechanism, a hardware adaptation system for the sensor is constructed.

[0028] The operating temperature range refers to the temperature range within which the sensor can maintain its basic functions (startup, measurement, and communication) without failure, and must cover the extreme temperatures of high-altitude and cold environments, such as -45℃ to +85℃. The operating atmospheric pressure range refers to the atmospheric pressure range within which the sensor can operate normally, corresponding to altitude variations, such as 50kPa to 110kPa. The low-temperature startup performance refers to the time and success rate required for the sensor to output stable data from power-on at its lowest operating temperature (e.g., -45℃). The low-temperature accuracy curve is a distribution chart describing the deviation between the measured value and the true value of the sensor across the entire temperature range (especially below -40℃). The low-temperature drift coefficient refers to the sensor output offset caused by a unit temperature change. The low-temperature selection criteria refer to the sensor's low-temperature performance thresholds established by combining the above parameters, such as, at -45℃, startup time ≤ 30 seconds, accuracy loss ≤ 1% of the full range, and drift coefficient ≤ 100ppm / ℃. The pressure resistance performance refers to the sensor's housing and internal components' ability to withstand external atmospheric pressure changes (e.g., rapid depressurization) without deformation, sealing failure, or electrical breakdown. The electrical clearance refers to the shortest distance between conductive components of the sensor in the air, which must meet the insulation strength requirements under low air pressure, such as ≥3mm at 50kPa according to IEC60664-1 standard. The creepage distance refers to the shortest path along the surface of the insulating material between conductive components, preventing leakage due to surface condensation under low air pressure, such as ≥5mm. The low-pressure selection standard refers to the threshold requirements of comprehensive air pressure-related parameters, such as passing a withstand pressure test of 50kPa / 72h, electrical clearance ≥3mm, and creepage distance ≥5mm. The hardware materials refer to a combination of materials that meet low-temperature toughness and low-pressure insulation requirements, such as aerospace-grade aluminum alloy, silicone rubber, and polytetrafluoroethylene. The protective structure refers to a complete set of physical protection mechanisms designed for sensors and related electronic components in high-altitude, cold-weather batteries. The electrical connection structure refers to the complete design in the high-altitude, cold-weather battery system used to realize the electrical signal transmission and physical connection between the sensor and the battery management system, controller, or other electronic units. The signal conditioning structure refers to a circuit design system in high-altitude, cold-weather battery systems that preprocesses, optimizes, and converts the raw electrical signals output by sensors. The thermal management mechanism refers to a complete hardware and control strategy system in high-altitude, cold-weather battery systems used to actively or passively regulate and maintain the operating temperature of the battery and sensor system within a reasonable range.

[0029] Optionally, the low-temperature start-up performance can be determined by the low-temperature environmental chamber test method, the low-temperature accuracy curve can be determined by the multi-point calibration test method, and the low-temperature drift coefficient can be determined by the temperature cycle drift test.

[0030] Optionally, the withstand voltage performance can be determined by a low-pressure breakdown test, the electrical clearance can be determined by finite element electric field simulation, such as using electric field simulation software like ANSYS Maxwell or COMSOL, and the creepage distance can be determined by surface insulation resistance testing.

[0031] Optionally, the hardware adaptation system can be constructed through multi-physics co-simulation, such as simulating the coupling effects of multiple physical fields such as structure, heat, electricity, and flow through multi-physics co-simulation to construct the hardware adaptation system of the sensor.

[0032] This invention, through the sensor selection criteria and hardware adaptation system, constructs a multimodal sensing network for the high-altitude, cold-weather battery, enabling high-fidelity, multi-dimensional environmental-battery coupling state perception, thus providing a data foundation for subsequent data analysis. The multimodal sensing network refers to an intelligent sensing system in high-altitude, cold-weather battery systems, constructed to comprehensively, accurately, and in real-time monitor the state of the battery and its operating environment. This system comprises various types of sensors and possesses multi-dimensional information acquisition, fusion, and collaborative operation capabilities.

[0033] As an embodiment of the present invention, the construction of the multimodal sensing network for the high-altitude and cold-weather battery based on the sensor selection criteria and the hardware adaptation system includes: Based on the sensor selection criteria and the hardware adaptation system, the multimodal sensors and sensor network accessories for the high-altitude and cold-weather battery are determined. Construct a network layout diagram of the sensor network accessories and the multimodal sensor; Determine the connection method and communication protocol between the sensor network accessory and the multimodal sensor; Based on the network layout diagram, the connection method, and the communication protocol, a multimodal sensing network for the high-altitude and cold-weather battery is constructed.

[0034] The multimodal sensors refer to a set of sensors deployed in high-altitude and cold-weather battery systems to comprehensively perceive the complex state of the battery and its operating environment, capable of monitoring information in different physical and chemical dimensions, such as electrical modal sensors, thermal modal sensors, and environmental modal sensors. The sensor network accessories refer to all auxiliary hardware and components used to support, connect, protect, and empower the multimodal sensors, enabling them to function normally and reliably transmit data. The network layout diagram is a graphical representation used to describe the distribution, connection relationships, data flow, and topology of multimodal sensors and sensor network accessories in physical space within a high-altitude and cold-weather battery system. The connection method refers to the specific method for achieving electrical connectivity and data transmission at the physical level between multimodal sensors and sensor network accessories, and between accessories themselves. The communication protocol refers to a set of rules and formats commonly followed when exchanging data between nodes (sensors, DAU, gateways) within the multimodal sensor network, and between the network and upper-layer systems (BMS, cloud platform).

[0035] Optionally, the network layout diagram can be constructed using physical space modeling methods, such as CAD modeling or spatial constraint analysis.

[0036] Optionally, the connection method can be determined through supervised learning, such as XGBoost or Random Forest, and the communication protocol can be determined through digital twin technology, such as using platforms like MATLAB / Simulink, ANSYS, or NS3 to establish a digital twin model of the physical environment, sensor network, and communication link. Through the digital twin model, the communication protocol between the sensor network accessories and the multimodal sensor can be determined.

[0037] S2. Collect multimodal data of the high-altitude and cold-weather battery through the multimodal sensing network to analyze the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress corresponding to the environmental characteristics. Based on the battery-environment coupling relationship, perform feature fusion on the multimodal data to obtain fused data.

[0038] This invention, through the multimodal sensing network, collects multimodal data from the high-altitude and cold-weather battery, enabling comprehensive real-time acquisition of relevant battery data and achieving comprehensive perception and accurate assessment of battery status. The multimodal data refers to a heterogeneous data set collected through the multimodal sensing network, originating from different types of sensors and reflecting multiple dimensions of the battery's and its operating environment's state and characteristics, such as environmental modal data, battery modal data, and operating condition modal data.

[0039] This invention, through analyzing the battery-environment coupling relationship between the high-altitude, cold-climate battery and the corresponding environmental stress, reveals the mechanism by which environmental stress affects battery performance, enabling precise performance analysis of the battery in complex environments. The battery-environment coupling relationship refers to the dynamic relationship of interaction, mutual influence, and mutual constraint between the battery system and its external environment across multiple dimensions, including physical, chemical, thermodynamic, and electrochemical dimensions.

[0040] As an embodiment of the present invention, the analysis of the battery-environment coupling relationship between the high-altitude, cold-climate battery and the environmental stress corresponding to the environmental characteristics includes: The multimodal data corresponding to the high-altitude and cold-weather battery are preprocessed to obtain preprocessed multimodal data. The preprocessed multimodal data is discretized to obtain discretized multimodal data; Extract the discrete multimodal features from the discrete multimodal data; Calculate the correlation coefficient between the discretized multimodal characteristics and the environmental stress; Based on the correlation coefficient, a battery-environment dynamic coupling equation is constructed for the high-altitude and cold-weather battery and the environmental stress. Based on the battery-environment dynamic coupling equation, the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress is determined.

[0041] The preprocessed multimodal data refers to a structured and analyzable dataset obtained after cleaning, aligning, standardizing, and denoising the collected raw multimodal data. The discretized multimodal data refers to data in which continuous preprocessed data is divided into discrete segments according to time, space, and numerical range. The discretized multimodal features refer to key indicators extracted from the discretized data that reflect the state of the battery or environment. The correlation coefficient is a numerical indicator that measures the degree of correlation between the discretized multimodal features and environmental stress. The battery-environment dynamic coupling equation is a mathematical model established based on the correlation analysis results, describing the dynamic interaction between battery state and environmental stress.

[0042] Optionally, the discretized multimodal data can be obtained through discretization methods, such as time discretization, numerical discretization, state discretization, etc.

[0043] Optionally, the correlation coefficient can be calculated using correlation analysis methods, such as Pearson correlation coefficient, Spearman rank correlation coefficient, etc.

[0044] Optionally, constructing the battery-environment dynamic coupling equation for the high-altitude, cold-climate battery and the environmental stress based on the correlation coefficient includes: Based on the correlation coefficient, a linear influence matrix of the environmental stress on the high-altitude and cold-weather battery is constructed. Define the comprehensive impact index of the environmental stress on the high-altitude and cold-weather battery; Based on the comprehensive impact index, the nonlinear coupling interaction matrix of the environmental stress on the high-altitude and cold-weather battery is analyzed. Based on the discretized multimodal characteristics corresponding to the high-altitude and cold-climate battery, the self-evolution coefficient matrix of the high-altitude and cold-climate battery is determined. Based on the linear influence matrix, the nonlinear coupling interaction matrix, and the self-evolution coefficient matrix, a battery-environment dynamic coupling equation for the high-altitude and cold-climate battery and the environmental stress is constructed.

[0045] The linear influence matrix refers to the matrix representation of the linear influence relationship between environmental stresses (such as temperature, air pressure, radiation, etc.) and battery state variables (such as voltage, internal resistance, temperature, etc.). The comprehensive influence index is a comprehensive quantitative indicator used to assess the overall influence of the environment on battery performance. The nonlinear coupling interaction matrix describes the nonlinear interaction between environmental stresses and battery state variables. The self-evolution coefficient matrix refers to the internal laws governing the natural evolution of the battery's state variables over time without external environmental influences.

[0046] Optionally, the linear influence matrix of the environmental stress on the high-altitude and cold-weather battery can be constructed by multiple linear regression.

[0047] Optionally, the comprehensive impact index of environmental stress on the high-altitude and cold-weather battery can be defined by machine learning model output methods, such as random forests, neural networks, etc.

[0048] As another implementation, the battery-environment dynamic coupling equation is expressed by the following formula:

[0049] in, The battery-environment dynamic coupling equation represents the battery state and environmental stress corresponding to batteries in high-altitude and cold regions. This represents the battery state matrix of batteries used in high-altitude and cold regions. The environmental stress matrix represents the environmental stress. This represents the matrix of its own evolution coefficients. Represents the linear influence matrix. Represents the nonlinear coupling interaction matrix. This indicates the coupling operator. This represents the coupling interaction term between the battery state matrix and the environmental stress matrix.

[0050] It needs to be explained that in this application, the formula... This represents the self-evolution coefficient matrix, with its elements ranging from [-0.1, 0]. The larger the value, the slower the battery's self-state decay rate. The linear influence matrix has elements ranging from [-1, 1]. When an element is greater than 0, it indicates that the environmental stress has a promoting effect on the battery state. When an element is equal to 0, it indicates that the environmental stress has no direct effect on the battery state. When an element is less than 0, it indicates that the environmental stress has an inhibiting effect on the battery state. This represents a nonlinear coupling interaction matrix, with its elements ranging from [-1, 1]. When an element is greater than 0, it indicates that there is a synergistic promoting relationship between the environmental stress and the corresponding battery state. When an element is equal to 0, it indicates that there is no coupling relationship between the environmental stress and the corresponding battery state. When an element is less than 0, it indicates that there is a mutual inhibiting relationship between the environmental stress and the corresponding battery state.

[0051] This invention, through feature fusion of the multimodal data based on the battery-environment coupling relationship, yields fused data that can compensate for the shortcomings of a single modality, significantly improving the accuracy of predictive analysis and thus enhancing data quality and information integrity. The fused data refers to the dataset obtained by integrating and processing raw data from different modalities.

[0052] S3. Based on the fused data, calculate the ionization degree reduction coefficient, lithium ion mobility, and internal resistance change rate of the high-altitude and cold-weather battery to analyze the battery charging state of the high-altitude and cold-weather battery.

[0053] This invention, through the fusion data, calculates the ionization degree decrease coefficient, lithium-ion mobility, and internal resistance change rate of the high-altitude, cold-weather battery, thereby accurately quantifying the internal state evolution of the battery and enhancing battery performance analysis and health assessment capabilities. The ionization degree decrease coefficient is a dimensionless parameter describing the degree of ionization capacity decay of the battery electrolyte under low temperature, low pressure, and extreme environments. The lithium-ion mobility refers to the average drift velocity of lithium ions in the electrolyte under a unit electric field strength. The internal resistance change rate refers to the degree of change in the battery's internal resistance relative to its initial or standard state under specific environmental stresses (such as high altitude, cold weather) and during cyclic use.

[0054] As an embodiment of the present invention, the calculation of the ionization degree reduction coefficient, lithium-ion mobility, and internal resistance change rate of the high-altitude and cold-weather battery based on the fused data includes: Extract the temperature sequence, air pressure sequence, electrolyte concentration sequence, electrolyte viscosity, and electrochemical characteristics from the fused data; The real-time degree of ionization of the high-altitude, cold-weather battery is calculated based on the temperature sequence, the air pressure sequence, and the electrolyte concentration sequence. Based on the real-time degree of ionization, the degree of ionization reduction coefficient of the high-altitude and cold-weather battery is determined; Determine the lithium-ion diffusion coefficient of the high-altitude, cold-weather battery; The lithium-ion mobility of the high-altitude and cold-weather battery is calculated based on the electrolyte viscosity and the lithium-ion diffusion system. Based on the electrochemical characteristics, the real-time internal resistance of the high-altitude and cold-weather battery was calculated. The rate of change of internal resistance of the high-altitude and cold-weather battery is determined based on the real-time internal resistance.

[0055] The temperature sequence refers to the sequence of continuously collected internal temperature data of the battery within a specific time period. The pressure sequence refers to the sequence of atmospheric pressure data of the environment in which the battery is located within a specific time period. The electrolyte concentration sequence refers to the sequence of changes in the molar concentration of lithium salt (such as LiPF6) in the battery electrolyte over time. The electrolyte viscosity refers to the flow resistance of the electrolyte at specific temperatures and pressures. The electrochemical characteristics refer to the electrochemical behavior characteristics exhibited by the battery during charging and discharging, such as voltage, current, and AC impedance spectroscopy. The real-time ionization degree refers to the degree of dissociation of lithium salt in the electrolyte under the current temperature, pressure, and electrolyte concentration conditions. The lithium-ion diffusion rate refers to the diffusion rate of lithium ions in the electrolyte or electrode material. The real-time internal resistance refers to the total resistance of the battery under specific operating conditions.

[0056] Optionally, the lithium-ion diffusion coefficient of the high-altitude, cold-weather battery can be determined by electrochemical impedance spectroscopy, such as by applying a small sinusoidal AC voltage signal to the battery, measuring its current response, and obtaining a Nyquist plot. The Warburg impedance, which is related to the diffusion process, can then be extracted to calculate the diffusion coefficient.

[0057] Optionally, the real-time internal resistance of the high-altitude, cold-weather battery can be calculated using extended Kalman filtering and dual Kalman filtering.

[0058] As another implementation, the real-time degree of ionization is calculated using the following formula:

[0059] in, Indicates in Real-time ionization at any given moment Indicates the degree of ionization under standard conditions. Indicates An exponential function with base 0. This represents the ionization activation energy of a battery operating at high altitudes and in cold regions (unit: J / mol). This represents the gas constant (unit: J / (mol·K)). Indicating temperature series in Real-time temperature at any given moment (unit: K). Indicates standard temperature (unit: K). Indicating the pressure sequence in Real-time air pressure at any given moment (unit: Pa). This indicates standard atmospheric pressure (unit: Pa). This represents the air pressure influence coefficient.

[0060] It should be explained that, in the formula, the ionization activation energy... The energy barrier required to transform an ion from its stable state to its ionized state is typically 45,000 J / mol, obtained by fitting experimental data to the Arrhenius equation. The gas constant refers to the energy change per mole of gas per unit temperature change, with a value of 8.314 J / (mol·K). The pressure influence coefficient is a quantitative parameter characterizing the effect of pressure changes on battery performance, obtained through molecular dynamics simulations or experimental calibration. Its value ranges from -1 to 1. When the pressure influence coefficient is greater than 0, increased pressure inhibits ionization in high-altitude, cold-weather batteries; when it is less than 0, increased pressure promotes ionization. For example, it is calibrated to -0.15 at an altitude of 5000 meters.

[0061] As another implementation, the lithium-ion mobility is calculated using the following formula:

[0062] in, Indicates in Lithium-ion mobility at time t, Indicates the initial lithium-ion mobility. Indicates An exponential function with base 0. This represents the migration activation energy (unit: eV) of batteries operating at high altitudes and in cold regions. This represents the Boltzmann constant (unit: eV / K). Indicating temperature series in Real-time temperature at any given moment (unit: K). Indicates the overpotential influence coefficient. Indicates in The change in overpotential at any given time.

[0063] It should be explained that the migration activation energy refers to the energy barrier that a lithium ion needs to overcome to migrate from one position to an adjacent position in the electrolyte. This energy can be obtained through molecular dynamics simulations, and its typical range is 0.3 eV to 0.6 eV. In this application, it was determined through experimental fitting to be... =0.45eV. The Boltzmann constant is a constant used to describe the relationship between temperature and energy, with a value of... eV / K. The overpotential change refers to the difference between the actual potential and the equilibrium potential of the electrode reaction, which can be calculated using real-time voltage and current data. The potential influence coefficient is a parameter that quantifies the impact of overpotential change on lithium-ion mobility. This coefficient is determined through linear regression analysis of historical charge-discharge data and corresponding mobility changes, and its value ranges from [0.05, 0.2]. The larger the value, the stronger the influence of overpotential on mobility. In this application, the regression analysis yields... =0.12.

[0064] This invention, through analysis of the state of charge (SOC) of the battery at high altitudes and cold environments, comprehensively considers electrochemical and environmental influences, improving the accuracy of SOC estimation under low temperature and low voltage conditions. It also analyzes the risk of lithium deposition at low temperatures, preventing overcharging-induced safety accidents, and dynamically adjusts charging current and voltage to improve charging efficiency and battery life. The SOC refers to a quantitative assessment of the battery's current comprehensive capability for safe and efficient charging under the specific environmental stress of high altitudes and cold environments.

[0065] As an embodiment of the present invention, the analysis of the state of charge of the high-altitude and cold-weather battery includes: Based on the lithium-ion mobility and internal resistance change rate of the high-altitude and cold-weather battery, the chargeability index of the high-altitude and cold-weather battery is calculated. The temperature compensation factor of the high-altitude and cold-weather battery is determined based on the ionization degree reduction coefficient of the battery. Based on the temperature compensation factor and the lithium-ion mobility, the lithium plating risk coefficient of the high-altitude and cold-weather battery is calculated. Calculate the temperature rise rate of the high-altitude, cold-weather battery to analyze the thermal runaway risk coefficient of the high-altitude, cold-weather battery; The total risk coefficient of the high-altitude and cold-weather battery is determined based on the lithium plating risk coefficient and the thermal runaway risk coefficient. The state of charge of the high-altitude and cold-weather battery is determined based on the charging acceptability index and the total risk coefficient.

[0066] The charge acceptability index quantifies the battery's ability to safely accept a certain charging current under its current internal kinetic state. The temperature compensation factor corrects for the impact of temperature on the battery's electrochemical reactions. The lithium plating risk coefficient represents the probability of lithium metal plating occurring under low-temperature, high-rate charging conditions. The thermal runaway risk coefficient assesses the risk of thermal runaway caused by excessively rapid temperature rise. The overall risk coefficient combines the risks of lithium plating and thermal runaway, reflecting the overall risk level of the battery under its current charging state.

[0067] Optionally, the charge acceptability index of the high-altitude and cold-weather battery can be calculated by a multi-parameter dynamic weighted fusion method. This method regards lithium-ion mobility and internal resistance change rate as dynamically changing electrochemical parameters. Through real-time weight adjustment, it reflects the degree of their influence under different temperature, air pressure and current conditions, thereby calculating the charge acceptability index.

[0068] Optionally, the temperature compensation factor of the high-altitude and cold-weather battery can be determined by an adaptive temperature compensation method based on fuzzy logic. This method uses a fuzzy logic system, takes the ionization degree reduction coefficient and ambient temperature and air pressure as inputs, and outputs an adaptive temperature compensation factor.

[0069] Optionally, the lithium plating risk coefficient of the high-altitude and cold-weather battery can be calculated by electrochemical-thermal coupling modeling. For example, by using an electrochemical-thermal coupling model, factors such as temperature, lithium-ion mobility, and ionization degree reduction coefficient are comprehensively considered to simulate the electrochemical reaction and heat distribution inside the battery, thereby calculating the lithium plating risk coefficient.

[0070] S4. Define the charging target of the high-altitude and cold-weather battery, construct the target excitation function of the high-altitude and cold-weather battery, define the control action set of the high-altitude and cold-weather battery, integrate the control action set and the target excitation function into a preset reinforcement learning framework, and obtain the charging strategy decision network.

[0071] This invention provides a precise criterion for determining the charging target of the high-altitude, cold-weather battery, thus offering a data foundation for subsequent target optimization. The charging target refers to the optimization control objectives set for the battery system in high-altitude, cold-weather environments, such as safety targets, efficiency targets, and environmental adaptability targets.

[0072] This invention, through the construction of a target activation function for the high-altitude, cold-weather battery, can balance multiple conflicting objectives, achieving autonomous optimization and continuous learning, thereby improving the charging system's adaptability to extreme operating conditions. The target activation function, within the reinforcement learning framework for high-altitude, cold-weather batteries, is a mathematical function used to comprehensively evaluate the gains obtained by the agent in executing charging control strategies across multiple key objective dimensions.

[0073] As an embodiment of the present invention, the construction of the target excitation function for the high-altitude, cold-weather battery includes: The charging target corresponding to the high-altitude and cold-weather battery is decomposed into multiple sub-targets; Construct multiple incentive functions for the multiple sub-objectives, wherein the multiple incentive functions include: a safety reward function, an efficiency reward function, and a lifetime reward function; Determine the importance weights of the aforementioned multiple activation functions; Based on the importance weights, the multiple activation functions are weighted and summed to obtain the target activation function.

[0074] The term "multiple sub-objectives" refers to decomposing the charging objectives that need to be optimized during the charging process of high-altitude and cold-climate battery systems into several specific, quantifiable, and independent sub-objectives, such as safety sub-objectives (e.g., avoiding dangerous states such as lithium plating, thermal runaway, overcharging, and over-discharging), efficiency sub-objectives (e.g., improving charging speed, reducing energy loss, and improving charging efficiency), and lifespan sub-objectives (e.g., slowing capacity decay, suppressing internal resistance growth, and extending battery cycle life). The term "multiple incentive functions" refers to reward functions designed for each of the above sub-objectives, used to quantify the immediate feedback obtained by the intelligent agent in each sub-objective dimension after taking action under specific conditions. The term "safety reward function" refers to incentive functions used to assess whether the battery is in a safe state during charging, such as lithium plating risk reward functions and voltage anomaly reward functions. The term "efficiency reward function" refers to incentive functions used to assess the efficiency of the charging process, such as charging time reward functions and energy efficiency reward functions. The term "lifespan reward function" refers to incentive functions used to assess the maintenance of battery lifespan, such as capacity decay reward functions and internal resistance growth reward functions. The importance weight refers to the weight coefficient assigned to each activation function to reflect the relative importance of different sub-objectives in the overall optimization.

[0075] Optionally, the multiple incentive functions can be constructed using a hybrid approach of fuzzy logic and reward shaping. This approach combines fuzzy logic to shape the reward, resulting in multiple incentive functions that make the reward function smoother, more instructive, and accelerate convergence.

[0076] Optionally, the importance weights can be determined by dynamic weight allocation of the attention mechanism.

[0077] In another implementation, the target excitation function is calculated using the following formula:

[0078] in, Represents the target activation function. This represents the security reward function. This represents the efficiency reward function. Represents the lifetime reward function, This represents the weight of the security reward function. The weights represent the efficiency reward function. This represents the weight of the lifetime reward function.

[0079] It needs to be explained that in this application, This represents the weight of the security reward function, with a value range of (0,1). The weights of the efficiency reward function are represented, and their values ​​range from (0,1). The weights of the lifetime reward function are denoted by , and their values ​​range from (0,1) and satisfy . .

[0080] This invention, through defining a set of control actions for the high-altitude, cold-weather battery, allows for the selection of the optimal control strategy based on different environmental and battery states, achieving refined control of the charging process. The set of control actions refers to the set of all possible operations that the agent can choose to execute at each time step within the reinforcement learning framework for the high-altitude, cold-weather battery, based on the current state. Examples include charging current adjustment (such as setting different levels like 0.1C, 0.2C, 0.5C, 1C, or continuously adjusting the current value), charging voltage adjustment (such as setting target voltages like 3.7V, 4.0V, 4.2V, etc.), and temperature control actions (such as turning the heater on / off and adjusting the heating power (e.g., 10W, 20W, 30W)).

[0081] Optionally, the set of control actions can be defined using a multimodal action space design method, such as constructing an action space that supports multiple control modes (e.g., current control, voltage control, temperature control) corresponding to the reinforcement learning framework, in order to define the set of control actions for the reinforcement learning framework.

[0082] This invention integrates the set of control actions and the target activation function into a preset reinforcement learning framework to obtain a charging strategy decision network that can achieve multi-objective collaborative optimization, improving the intelligence and adaptability of charging control. The preset reinforcement learning framework is a comprehensive technical architecture integrating an environment model, an agent, a learning mechanism, a decision network, and evaluation criteria. The charging strategy decision network is an intelligent decision-making system that, in high-altitude and cold environments, autonomously selects the optimal charging control action based on the current battery state and environmental conditions to achieve comprehensive optimization of multiple objectives such as safety, efficiency, and lifespan.

[0083] As an embodiment of the present invention, the integration of the control action set and the target activation function into a preset reinforcement learning framework to obtain a charging policy decision network includes: Obtain historical battery data for the high-altitude, cold-weather battery corresponding to the target excitation function; Based on the historical battery data, the electrochemical and thermodynamic characteristics of the high-altitude and cold-weather battery were analyzed. Based on the electrochemical and thermodynamic characteristics, a battery model for the high-altitude and cold-weather battery is constructed. Based on the environmental characteristics of the high-altitude and cold-weather battery, a high-altitude and cold-weather environmental model of the battery is constructed. The battery model and the high-altitude and cold environment model are integrated to obtain a battery environment simulation model. Based on the target activation function, a reinforcement learning algorithm for the high-altitude and cold-weather battery is determined. Based on the reinforcement learning algorithm and the set of control actions, the learning network structure of the high-altitude and cold-weather battery is constructed. By combining the battery environment simulation model, the learning network structure, and the reinforcement learning framework, a charging strategy decision network for the high-altitude and cold-weather battery is integrated.

[0084] The historical battery data refers to the collection of all relevant data generated during the target battery's past operation and testing, such as battery voltage, battery current, and state of charge, which can be obtained using battery testing equipment (such as Arbin or Bio-Logic testers). The electrochemical characteristics refer to the set of features describing the insertion and extraction processes of active materials (lithium ions) within the battery, phase transitions in electrode materials, and the resulting potential changes, such as open-circuit voltage, polarization voltage, and polarization resistance. The thermodynamic behavior characteristics refer to the set of features describing the heat exchange between the battery and the external environment during operation, as well as its internal heat generation and temperature distribution, such as heat generation rate, core temperature, and temperature gradient. The battery model refers to a mathematical model used to describe the internal electrochemical characteristics, thermodynamic behavior, and aging mechanisms of the battery under high-altitude and cold conditions. The high-altitude and cold environment model refers to a mathematical model describing the external environmental conditions of the battery, including parameters such as temperature, air pressure, air density, and wind speed, and their impact on battery thermal management and charge / discharge behavior. The battery environment simulation model refers to a system platform that couples the battery model with the environment model to form an interactive and simulateable system platform. This platform simulates the dynamic behavior of batteries in high-altitude and cold environments, providing a training and testing environment for reinforcement learning. The reinforcement learning algorithm refers to computational methods used to train agents in high-altitude and cold battery systems to learn optimal charging strategies, such as DQN, DDPG, and PPO. Specifically, the state space is designed as an 8-dimensional vector containing battery temperature, internal resistance, lithium plating risk coefficient, and thermal runaway risk coefficient. The action space is defined as charging current levels (e.g., 0.1C, 0.2C, 0.5C, 1C) and heating power levels (e.g., 0W, 10W, 20W). The learning network structure refers to a neural network structure used to approximate the target value function.

[0085] Optionally, the battery model can be constructed by coupling the electrochemical mechanism (such as the P2D model) and thermodynamic model of the high-altitude, cold-climate battery. The high-altitude, cold-climate environment model can be constructed using a meteorological-geographical coupling model, which combines meteorological data (temperature, humidity, wind speed) and geographic information (altitude, topography) to construct the high-altitude, cold-climate environment model of the battery.

[0086] Optionally, the learning network structure can be constructed using neural networks, such as multilayer perceptrons (MLP), convolutional neural networks (CNN), recurrent neural networks, etc.

[0087] S5. Based on the battery charging state, output the target charging command for the high-altitude and cold-weather battery through the charging strategy decision network, and execute the dynamic charging of the high-altitude and cold-weather battery based on the target charging command.

[0088] This invention, through its embodiment, enables intelligent and adaptive control of high-altitude and cold-weather charging by outputting target charging commands for the battery based on its charging state via a charging strategy decision network. This dynamically adjusts the weights of safety, efficiency, and lifespan at different stages, achieving multi-objective collaborative optimization and improving the adaptability of the charging strategy in extreme environments. The target charging command refers to the optimal control command generated by the charging strategy decision network based on the current battery state and environmental conditions, such as charging current commands, charging voltage commands, and temperature compensation commands.

[0089] The embodiments of the present invention can improve charging safety and efficiency, and enhance the environmental adaptability of charging in high-altitude and cold regions by performing dynamic charging of the battery based on the target charging command.

[0090] Optionally, to verify the effectiveness of the method described in this invention, a simulation comparison experiment was conducted in a high-altitude and cold environment. By simulating the extreme working conditions of the battery at an altitude of 5000 meters (air pressure of approximately 54 kPa) and a temperature of -40°C, the simulation results are shown in Table 1.

[0091] As can be clearly seen from Table 1 above, under the aforementioned extreme conditions, compared with traditional methods, the charging method provided by this invention can significantly improve charging efficiency by approximately 20 percentage points (a relative improvement of 29.4%), effectively reduce energy loss, lower the lithium plating risk factor by 40%, greatly enhance the safety of the charging process, shorten the total charging time by 24%, and improve the charging speed.

[0092] Compared to the problems described in the background technology, this invention provides a scientific basis for sensor selection and hardware adaptation through in-depth analysis of the characteristics of high-altitude and cold environments. Complex environmental stresses such as low temperature, low air pressure, and strong radiation directly affect the electrochemical reaction kinetics and material stability of batteries. Through systematic research on these environmental characteristics, this invention can accurately identify key parameters affecting battery performance, such as temperature gradient, air pressure changes, and radiation intensity. This not only guides the selection of high-precision, high-reliability sensors, ensuring the accuracy and real-time nature of data acquisition, but also promotes the specialized design of hardware systems, such as low-temperature start-up circuits and air pressure compensation structures, fundamentally improving the adaptability and robustness of the BMS (Battery Management System) in extreme environments. Secondly, based on multimodal sensor networks and feature fusion technology, this invention achieves deep perception and precise characterization of battery status. By constructing a multimodal sensor network, this invention integrates multidimensional data such as electrical, thermal, mechanical, and gaseous data, and performs feature-level fusion based on a battery-environment coupling relationship model, effectively eliminating the limitations of a single sensor. This invention further calculates key internal parameters such as the ionization degree decrease coefficient, lithium-ion mobility, and internal resistance change rate, elevating the battery charging state to a comprehensive state vector encompassing electrochemical characteristics, thermodynamic behavior, and safety boundaries. This deep perception capability provides a high-quality, high-information-density data foundation for subsequent intelligent decision-making. Finally, the introduction of a reinforcement learning decision network evolves the charging strategy from fixed rules to dynamic intelligence, achieving optimal control under multi-objective optimization. This invention constructs charging objectives (such as speed, safety, and lifespan) as objective excitation functions and defines parameters such as charging current and voltage as a set of control actions. By using the fused battery state as input, the reinforcement learning network can evaluate the long-term rewards of different charging actions in real time and output the target charging command under the current operating condition. This dynamic charging strategy can intelligently balance fast charging and safety protection based on the actual battery state and environmental changes, effectively avoiding risks such as low-temperature lithium plating and high-voltage overcharging, significantly shortening charging time, and maximizing battery cycle life. Therefore, the high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning provided in this embodiment of the invention can improve the safety and efficiency of high-altitude and cold-weather battery charging.

[0093] like Figure 2 The diagram shown is a functional block diagram of a high-altitude and cold-weather battery charging system based on multimodal perception and reinforcement learning according to the present invention.

[0094] The high-altitude, cold-weather battery charging system 200 based on multimodal perception and reinforcement learning described in this invention can be installed in an electronic device. Depending on the functions implemented, the high-altitude, cold-weather battery charging system based on multimodal perception and reinforcement learning may include a multimodal sensor network construction module 201, a data fusion module 202, a battery charging state analysis module 203, a charging strategy decision network construction module 204, and a charging execution module 205. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, stored in the memory of the electronic device.

[0095] In this embodiment of the invention, the functions of each module / unit are as follows: The multimodal sensor network construction module 201 is used to analyze the environmental characteristics of the high-altitude and cold-weather battery, to determine the sensor selection criteria and hardware adaptation system of the high-altitude and cold-weather battery, and to construct the multimodal sensor network of the high-altitude and cold-weather battery based on the sensor selection criteria and the hardware adaptation system. The data fusion module 202 is used to collect multimodal data of the high-altitude and cold-weather battery through the multimodal sensing network, so as to analyze the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress corresponding to the environmental characteristics, and perform feature fusion on the multimodal data based on the battery-environment coupling relationship to obtain fused data. The battery charging state analysis module 203 is used to calculate the ionization degree reduction coefficient, lithium ion mobility and internal resistance change rate of the high-altitude and cold-weather battery based on the fused data, so as to analyze the battery charging state of the high-altitude and cold-weather battery. The charging strategy decision network construction module 204 is used to clarify the charging target of the high-altitude and cold-weather battery, construct the target excitation function of the high-altitude and cold-weather battery, define the control action set of the high-altitude and cold-weather battery, and integrate the control action set and the target excitation function into a preset reinforcement learning framework to obtain the charging strategy decision network. The charging execution module 205 is used to output a target charging command for the high-altitude and cold-weather battery through the charging strategy decision network based on the battery charging state, and to perform dynamic charging of the high-altitude and cold-weather battery based on the target charging command.

[0096] In detail, the modules in the high-altitude and cold-weather battery charging system 200 based on multimodal perception and reinforcement learning described in this embodiment of the invention employ the same methods as described above. Figure 1 The method described above is the same as the high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning, and can produce the same technical effect, so it will not be repeated here.

[0097] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0098] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for charging batteries in high-altitude and cold regions based on multimodal perception and reinforcement learning, characterized in that, The method includes: The environmental characteristics of the high-altitude and cold-weather battery are analyzed to determine the sensor selection criteria and hardware adaptation system for the battery. Based on the sensor selection criteria and hardware adaptation system, a multimodal sensing network for the high-altitude and cold-weather battery is constructed. Multimodal data of the high-altitude and cold-weather battery are collected through the multimodal sensing network to analyze the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress corresponding to the environmental characteristics. Based on the battery-environment coupling relationship, feature fusion is performed on the multimodal data to obtain fused data. Based on the fused data, the ionization degree reduction coefficient, lithium ion mobility, and internal resistance change rate of the high-altitude and cold-weather battery are calculated to analyze the battery charging state of the high-altitude and cold-weather battery. The charging objective of the high-altitude and cold-weather battery is defined to construct the target excitation function of the high-altitude and cold-weather battery. The control action set of the high-altitude and cold-weather battery is defined, and the control action set and the target excitation function are integrated into a preset reinforcement learning framework to obtain the charging strategy decision network. Based on the battery charging state, the target charging command for the high-altitude and cold-weather battery is output through the charging strategy decision network, and dynamic charging of the high-altitude and cold-weather battery is executed based on the target charging command.

2. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The calculation of the ionization degree reduction coefficient, lithium-ion mobility, and internal resistance change rate of the high-altitude and cold-weather battery based on the fused data includes: Extract the temperature sequence, air pressure sequence, electrolyte concentration sequence, electrolyte viscosity, and electrochemical characteristics from the fused data; The real-time degree of ionization of the high-altitude, cold-weather battery is calculated based on the temperature sequence, the air pressure sequence, and the electrolyte concentration sequence. Based on the real-time degree of ionization, the degree of ionization reduction coefficient of the high-altitude and cold-weather battery is determined; Determine the lithium-ion diffusion coefficient of the high-altitude, cold-weather battery; The lithium-ion mobility of the high-altitude and cold-weather battery is calculated based on the electrolyte viscosity and the lithium-ion diffusion system. Based on the electrochemical characteristics, the real-time internal resistance of the high-altitude and cold-weather battery was calculated. The rate of change of internal resistance of the high-altitude and cold-weather battery is determined based on the real-time internal resistance.

3. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The analysis of the state of charge of the high-altitude, cold-weather battery includes: Based on the lithium-ion mobility and internal resistance change rate of the high-altitude and cold-weather battery, the chargeability index of the high-altitude and cold-weather battery is calculated. The temperature compensation factor of the high-altitude and cold-weather battery is determined based on the ionization degree reduction coefficient of the battery. Based on the temperature compensation factor and the lithium-ion mobility, the lithium plating risk coefficient of the high-altitude and cold-weather battery is calculated. Calculate the temperature rise rate of the high-altitude, cold-weather battery to analyze the thermal runaway risk coefficient of the high-altitude, cold-weather battery; The total risk coefficient of the high-altitude and cold-weather battery is determined based on the lithium plating risk coefficient and the thermal runaway risk coefficient. The state of charge of the high-altitude and cold-weather battery is determined based on the charging acceptability index and the total risk coefficient.

4. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The analysis of the battery-environment coupling relationship between the high-altitude, cold-climate battery and the environmental stress corresponding to the environmental characteristics includes: The multimodal data corresponding to the high-altitude and cold-weather battery are preprocessed to obtain preprocessed multimodal data. The preprocessed multimodal data is discretized to obtain discretized multimodal data; Extract the discrete multimodal features from the discrete multimodal data; Calculate the correlation coefficient between the discretized multimodal characteristics and the environmental stress; Based on the correlation coefficient, a battery-environment dynamic coupling equation is constructed for the high-altitude and cold-weather battery and the environmental stress. Based on the battery-environment dynamic coupling equation, the battery-environment coupling relationship between the high-altitude and cold-weather battery and the environmental stress is determined.

5. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 4, characterized in that, The process of constructing the battery-environment dynamic coupling equation based on the correlation coefficient includes: Based on the correlation coefficient, a linear influence matrix of the environmental stress on the high-altitude and cold-weather battery is constructed. Define the comprehensive impact index of the environmental stress on the high-altitude and cold-weather battery; Based on the comprehensive impact index, the nonlinear coupling interaction matrix of the environmental stress on the high-altitude and cold-weather battery is analyzed. Based on the discretized multimodal characteristics corresponding to the high-altitude and cold-climate battery, the self-evolution coefficient matrix of the high-altitude and cold-climate battery is determined. Based on the linear influence matrix, the nonlinear coupling interaction matrix, and the self-evolution coefficient matrix, a battery-environment dynamic coupling equation for the high-altitude and cold-climate battery and the environmental stress is constructed.

6. The high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The analysis of the environmental characteristics of batteries in high-altitude and cold regions includes: Clearly define the application scenarios for the high-altitude and cold-weather batteries; Determine the defining geographical and climatic conditions of the application scenario; Based on the defined geographical and climatic conditions, the environmental factors of the high-altitude and cold-climate battery are determined. Based on the aforementioned environmental factors, the environmental stress of the high-altitude and cold-weather battery is determined, wherein the environmental stress includes: primary environmental stress and secondary environmental stress; Based on the environmental stress, the environmental characteristics of the high-altitude and cold-weather battery are determined.

7. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The process of determining the sensor selection criteria and hardware adaptation system for the high-altitude and cold-weather battery includes: Based on the environmental characteristics corresponding to the high-altitude and cold-weather battery, the operating temperature range and operating air pressure range of the sensor corresponding to the high-altitude and cold-weather battery are determined. Based on the operating temperature range, the low-temperature start-up performance, low-temperature accuracy curve, and low-temperature drift coefficient of the sensor are determined. Based on the low-temperature start-up performance, the low-temperature accuracy curve, and the low-temperature drift coefficient, the low-temperature selection criteria for the sensor are determined. Based on the operating air pressure range, the pressure resistance, electrical clearance, and creepage distance of the sensor are determined. Based on the pressure resistance, the electrical clearance, and the creepage distance, the low-pressure selection criteria for the sensor are determined. Based on the low-temperature selection criteria and the air pressure selection criteria, the sensor selection criteria for the high-altitude and cold-weather battery are determined. The hardware materials of the sensor are determined according to the sensor selection criteria. Based on the aforementioned hardware materials, the sensor's protective structure, electrical connection structure, signal conditioning structure, and thermal management mechanism are constructed. Based on the protective structure, the electrical connection structure, the signal conditioning structure, and the thermal management mechanism, a hardware adaptation system for the sensor is constructed.

8. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The construction of the multimodal sensing network for the high-altitude, cold-weather battery based on the sensor selection criteria and the hardware adaptation system includes: Based on the sensor selection criteria and the hardware adaptation system, the multimodal sensor and sensor network accessories for the high-altitude and cold-weather battery are determined. Construct a network layout diagram of the sensor network accessories and the multimodal sensor; Determine the connection method and communication protocol between the sensor network accessory and the multimodal sensor; Based on the network layout diagram, the connection method, and the communication protocol, a multimodal sensing network for the high-altitude and cold-weather battery is constructed.

9. The high-altitude and cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The process of integrating the set of control actions and the target activation function into a preset reinforcement learning framework to obtain a charging policy decision network includes: Obtain historical battery data for the high-altitude, cold-weather battery corresponding to the target excitation function; Based on the historical battery data, the electrochemical and thermodynamic characteristics of the high-altitude and cold-weather battery were analyzed. Based on the electrochemical and thermodynamic characteristics, a battery model for the high-altitude and cold-weather battery is constructed. Based on the environmental characteristics of the high-altitude and cold-weather battery, a high-altitude and cold-weather environmental model of the battery is constructed. The battery model and the high-altitude and cold environment model are integrated to obtain a battery environment simulation model. Based on the target activation function, a reinforcement learning algorithm for the high-altitude and cold-weather battery is determined. Based on the reinforcement learning algorithm and the set of control actions, the learning network structure of the high-altitude and cold-weather battery is constructed. By combining the battery environment simulation model, the learning network structure, and the reinforcement learning framework, a charging strategy decision network for the high-altitude and cold-weather battery is integrated.

10. The high-altitude, cold-weather battery charging method based on multimodal perception and reinforcement learning as described in claim 1, characterized in that, The target excitation function for constructing the high-altitude, cold-weather battery includes: The charging target corresponding to the high-altitude and cold-weather battery is decomposed into multiple sub-targets; Construct multiple incentive functions for the multiple sub-objectives, wherein the multiple incentive functions include: a safety reward function, an efficiency reward function, and a lifetime reward function; Determine the importance weights of the aforementioned multiple activation functions; Based on the importance weights, the multiple activation functions are weighted and summed to obtain the target activation function.

Images