Intelligent compatible and efficient charging control system and method and cloud platform management system

Abstract

According to the intelligent compatible and efficient charging control system and the application method thereof, a management system combined with a cloud platform is provided, and a more intelligent, convenient and reliable charging control system is provided for a user by integrating the latest information technology, automation technology and artificial intelligence technology. Charging strategies suitable for various vehicle types are dynamically generated through the cloud strategy management platform, the edge computing node executes a lightweight charging control model, the multi-mode sensing module collects battery state data in real time, and the federal learning framework is connected with vehicle enterprises, power grid operators and charging facility data. The system solves the problems of poor compatibility and insufficient data collaboration of a traditional charging system, improves the charging efficiency, guarantees the safety of a battery, achieves the personalized adjustment of a charging strategy, and will show huge market potential and social value in the future along with the rapid growth of the intelligent equipment and electric vehicle market.

Description

Technical Field

[0001] The present application relates to the field of battery charging management, and more specifically, to an intelligent, compatible, and efficient charging control system and implementation method. Background Art

[0002] Current charging management systems for new energy vehicles face numerous challenges, primarily compatibility issues caused by the diversity of vehicle models, safety risks arising from complex battery structures, and the lack of a systematic charging strategy learning mechanism. Traditional charging systems generally use universal charging templates with fixed parameters, which are unable to adapt to the varying electrochemical characteristics of batteries in various vehicle models. Charging new models often requires redesigning charging curves, resulting in low charging efficiency and associated safety risks. For example, invention patent CN110635183A discloses a power battery system and a charging method with an optimized heating strategy for low-temperature charging. By implementing a heating and charging strategy with multi-temperature zone control, this method effectively reduces charging time in low-temperature environments for electric vehicle power battery systems, avoids frequent activation of the heating system, and significantly improves the user experience. However, this method primarily focuses on improving heating efficiency and fails to address energy recovery technology. From a grid interaction perspective, current charging facilities suffer from severe information silos between grid operators and automakers, significantly hindering the real-time dynamic adjustment of charging strategies. Furthermore, issues such as reduced battery performance in low-temperature conditions, the risk of thermal runaway caused by rapid charging, and the need for real-time monitoring of battery aging have all hindered the development of more intelligent charging systems. There is an urgent need for a comprehensive solution that can fully address multi-vehicle compatibility, dynamic strategy optimization, multi-dimensional safety monitoring and multi-agent data sharing. Summary of the Invention

[0003] In order to solve the problems and defects of the background technology, the present invention provides an intelligent, compatible and efficient charging control system, method and corresponding cloud platform charging management system, which are mainly aimed at electric vehicle charging control and management.

[0004] Intelligent compatible and efficient charging control system, characterized by including:

[0005] Connecting seat, used for connecting with external tram;

[0006] A central processing module, connected to the connection socket, for receiving and processing information including the charging power of the electric vehicle;

[0007] A locking module is provided between the central processing module and the connection socket, and is used to lock the connection after power matching;

[0008] An adaptation module, including a voltage regulation unit and a preset unit, for adjusting the voltage to match the tram requirements;

[0009] The control module is connected to the central processing module and the locking module and is used to adjust the locking state; the control module consists of a CAN bus, a voltage stabilizing unit and a control unit, wherein the CAN bus is connected to the voltage stabilizing unit and the control unit respectively, and its signal output end is connected to the locking module, and the control unit is connected to the central processing module.

[0010] The logic control module is located between the connector and the central processing module and is used to coordinate data interaction;

[0011] The intelligent algorithm module includes a core computing unit and a data storage unit. The intelligent algorithm module is connected to the central processing module and executes the instructions of the central processing module. It is used to analyze the historical usage data of the electric vehicle and predict the future power consumption pattern, thereby intelligently adjusting the charging strategy.

[0012] At the same time, a battery maintenance module is also provided, and the battery charging maintenance adopts a non-physical intelligent maintenance strategy and a multi-stage pulse repair method.

[0013] A multimodal sensing module is further provided, which is provided with multiple sensors, including a voltage sensor, a current sensor, a temperature sensor and a pressure sensor, for comprehensively monitoring the working status of the battery;

[0014] It further includes: a fiber optic tension sensor for monitoring the expansion deformation rate of the battery module; an electrochemical impedance spectroscopy acquisition unit for collecting frequency domain response signals at a time interval not exceeding 30 seconds; a non-contact infrared sensor array for real-time mapping of the temperature distribution on the battery surface; and the multimodal sensing module for real-time acquisition of battery voltage, current, temperature and expansion deformation data.

[0015] The charging method based on the above intelligent compatible and efficient charging control system is characterized in that:

[0016] Including connecting the tram to the connecting seat;

[0017] The central processing module detects the power of the tram and compares it with the preset parameters. Based on the comparison results, it decides whether to lock it directly or adjust the voltage first and then lock it;

[0018] Repeat the detection and adjustment until the electric vehicle can complete the entire charging process at constant power.

[0019] When the electric vehicle is initially connected to the connection socket, the central processing module sets the parameters of the preset unit according to the detected electric vehicle power; when it is detected that the connection socket power matches the preset parameters of the electric vehicle, the central processing module drives the locking module through the control module to achieve locking and start charging;

[0020] When it is detected that the power of the connection socket does not match the preset parameters of the tram, the central processing module also instructs the boost module in the adapter module to adjust the voltage through the control module until it meets the preset parameters, and then drives the locking module to execute the locking action again to start charging.

[0021] If the docking station power does not match the tram's preset parameters, the system will issue a warning through the display unit.

[0022] It also includes a battery health management method that uses a multimodal sensing module to monitor battery status.

[0023] The method comprises the following steps: S1, connecting the multimodal sensing module to the battery and starting monitoring;

[0024] S2. Pre-process the collected data through the data processing chip to eliminate noise interference;

[0025] S3. The central monitoring unit analyzes the pre-processed data to identify potential battery failures, such as overheating, overcharging, and internal short circuits, and records the time, type, and severity of the failure.

[0026] S4. For different fault types, the central monitoring unit initiates corresponding emergency response strategies, such as reducing the charge and discharge rate, starting the cooling system, disconnecting the circuit, etc.

[0027] S5. Generate battery health reports regularly, including battery life predictions, maintenance recommendations, etc., to help users take preventive measures in a timely manner and extend battery life.

[0028] It also includes battery maintenance methods; a core waveform generator is set in the circuit of the charging pile to provide a maintenance strategy based on the battery health report; and the charging waveform is intelligently controlled through the waveform pulse generation mechanism in the multi-stage pulse repair method to achieve battery maintenance.

[0029] A cloud platform management system based on intelligent compatibility and efficient charging control system, characterized by including:

[0030] The cloud-based policy management platform includes a data acquisition module, an analysis and processing module, and a policy generation module.

[0031] The data acquisition module is responsible for obtaining vehicle information and user demand information in real time from multiple vehicle management systems, user feedback channels and vehicle terminals;

[0032] The analysis and processing module performs comprehensive analysis and processing on the received information to determine whether the current state of the vehicle meets specific charging requirements, such as the remaining battery capacity, health status, and temperature.

[0033] The strategy generation module dynamically generates personalized charging strategies for different vehicle types based on the analysis and processing results;

[0034] The cloud-based policy management platform further includes:

[0035] A vehicle clustering engine creates a digital twin of the vehicle and optimizes charging curve characteristics based on the battery's electrochemical properties and charging graphs.

[0036] The reinforcement learning trainer generates a strategy component that optimizes charging efficiency, temperature rise suppression, and life extension. This strategy is implemented using a multi-objective reward function with dynamic weight adjustment. A gradient penalty mechanism is used to balance the constraints between different optimization objectives, ensuring that the reinforcement learning strategy remains within the battery's safe operating range.

[0037] Transfer learning module, which enables the migration of optimized strategy parameters of verified vehicle models to new battery systems;

[0038] Edge computing nodes, deployed in charging piles or vehicle-mounted ECUs, are used to execute lightweight charging control models;

[0039] A multimodal sensing module, which is used to collect data on battery voltage, current, temperature, and expansion deformation in real time;

[0040] A federated learning framework is used to connect data from automakers, power grid operators, and charging facilities.

[0041] The reinforcement learning trainer performs dynamic weight allocation based on a multi-objective reward function to generate a charging strategy that meets multi-dimensional constraints. It further maintains response delay and voltage and temperature joint control loops in automotive-grade chips based on the compression model of neural architecture search, and dynamically adjusts the constant current and constant voltage thresholds in the fast charging stage according to the real-time state of charge.

[0042] The multimodal sensing module integrates the technical solutions of distributed optical fiber strain sensors, electrochemical impedance spectroscopy acquisition units and non-contact infrared arrays. The data of the three types of sensors are synchronized through timestamps and input into the edge computing node to form a three-dimensional state matrix of mechanical deformation, electrochemical impedance and temperature distribution.

[0043] The vehicle digital twin model not only contains static data but also supports dynamic adjustment.

[0044] The beneficial effects of this patent are: the present invention provides an intelligent, compatible and efficient charging control system, method and cloud platform management system, which dynamically generates multi-vehicle charging strategies through a cloud-based strategy platform, real-time control of edge computing nodes, multi-modal sensor data collection and a federated learning framework to connect data islands, solving the problems of poor compatibility and insufficient data collaboration in traditional charging systems, improving charging efficiency, ensuring battery safety and promoting grid interaction, meeting the specific electrochemical requirements of batteries of different models, realizing personalized adjustment of charging strategies, and at the same time strengthening data interoperability between information islands, promoting the development of charging technology towards a smarter and safer direction. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of the working principle of intelligent compatible and efficient charging control system;

[0046] Figure 2 is based on Figure 1 Schematic diagram of the battery health management method;

[0047] Figure 3 A block diagram of the contents of generating a battery health report;

[0048] Figure 4 is based on Figure 1 Schematic diagram of the management system of the cloud platform; DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of embodiments of the present invention, rather than all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative work are within the scope of protection of the present application. It should be noted that the terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects and are not necessarily used to describe their order. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server comprising a series of steps or submodules is not necessarily limited to those steps or submodules clearly listed, but may include other steps or submodules that are not clearly listed or that are inherent to these processes, methods, products or devices.

[0050] Combined with attachment Figure 1 The present invention provides an intelligent, compatible and efficient charging control system, which includes but is not limited to: a connecting socket, a central processing module, a locking module, an adaption module, a regulation module, a logic control module, and a display module.

[0051] The connection socket is used to connect to an external electric vehicle; a central processing module, which is connected to the connection socket and is used to receive and process the electric vehicle charging power information transmitted from the connection socket; a locking module, which is provided between the central processing module and the connection socket and is used to lock the connection between the connection socket and the electric vehicle after matching is completed (locking parameters);

[0052] The adaptation module includes a voltage adjustment unit and a preset unit. The signal input end of the voltage adjustment unit is connected to the central processing module, and the output end is connected to the preset unit. The preset unit is connected to the connection socket signal. The central processing module determines whether the connection socket power matches the electric vehicle based on the received electric vehicle charging power information. If it matches, it is locked through the locking module. If it does not match, the central processing module adjusts the voltage adjustment unit to the preset unit's predetermined value and then locks it through the locking module.

[0053] The signal input end of the control module is connected to the central processing module, and the output end is connected to the locking module, which is used to adjust the locking state according to the instructions of the central processing module. The control module is specifically composed of a CAN bus, a voltage stabilizing unit and a control unit, wherein the CAN bus is respectively connected to the voltage stabilizing unit and the control unit, and its signal output end is connected to the locking module, and the control unit is connected to the central processing module. At the same time, a logic control module is provided, which is located between the connecting socket and the signal output end of the central processing module to coordinate the data interaction between the two. The adapter module has a built-in display module, the signal input end of the display module is connected to the voltage regulating unit and the preset unit, and the output end is connected to the connecting socket, and the current working status can be fed back in the form of indicator lights, display screens or buzzers, and the central processing module detects the actual power of the tram through NTC.

[0054] The system also innovatively incorporates intelligent algorithm modules, including a core computing unit and a data storage unit. The core computing unit utilizes a high-performance processor to ensure efficient execution of complex calculations, while the data storage unit utilizes a high-speed solid-state drive to ensure smooth reading and writing of large amounts of data. To enhance system reliability and security, a redundancy mechanism has been designed. If the primary computing unit or storage unit fails, a backup unit can immediately take over, ensuring uninterrupted service.

[0055] The intelligent algorithm module connects to the central processing module and executes its instructions. It analyzes the tram's historical usage data, predicts future power consumption patterns, and intelligently adjusts charging strategies to achieve optimal charging results. Specifically, when a user connects to a tram via the docking station, the central processing module first reads basic information about the tram, such as model and rated power, and immediately invokes the intelligent algorithm module. Based on stored historical data, the intelligent algorithm module uses machine learning algorithms, such as support vector machines (SVMs) or deep neural networks (DNNs), to assess the optimal charging parameters for the current tram.

[0056] The intelligent algorithm module in this embodiment is particularly suitable for large-scale electrical devices with multiple battery combinations, such as electric vehicles. In electric vehicle charging scenarios, the intelligent algorithm module can predict the vehicle's energy demand over a period of time based on the vehicle's driving history, such as common driving routes, average speed, and start frequency, and then dynamically adjust the charging mode. For example, if it predicts that the user is about to go on a long-distance trip, the charging speed will be accelerated in advance to ensure that the vehicle has sufficient power for long-distance driving. For daily urban commuting, a more economical low-speed charging mode is used to reduce electricity costs and extend battery life.

[0057] The present invention further provides a charging method based on the above-mentioned intelligent compatible and efficient charging control system, comprising connecting a tram to a connection socket; detecting the power of the tram through a central processing module and comparing it with preset parameters, and deciding whether to lock directly or adjust the voltage first and then lock according to the comparison result; repeating the detection and adjustment until the tram can complete the entire charging process at constant power. When the tram is initially connected to the connection socket, the central processing module sets the parameters of the preset unit according to the detected tram power; when it is detected that the power of the connection socket matches the preset parameters of the tram, the central processing module drives the locking module via the control module to achieve locking and start charging; when it is detected that the power of the connection socket does not match the preset parameters of the tram, the central processing module also adjusts the voltage through the boost module in the control module instruction adapter module until it meets the preset parameters, and then drives the locking module to perform the locking action again to start charging. If the power of the connection socket does not match the preset parameters of the tram, the system will issue a warning through the display unit. The present invention also relates to a charging system, which is mainly composed of the above-mentioned intelligent compatible and efficient charging circuit.

[0058] This system also takes into account the charging operation steps in low-temperature environments, such as running a heating mechanism stimulated by the resistance method to maintain the conductivity of the electrolyte solution; introducing a long-short memory network with physical constraints to correct the SOC estimation deviation caused by temperature; and flexibly defining a constant current charging cycle covering the range from 0% to 60% SOC.

[0059] The system also features fault prediction. By continuously monitoring the current and voltage during charging, the intelligent algorithm module can identify potential equipment anomalies and issue timely warnings to notify maintenance personnel for inspection. This feature helps prevent widespread power outages caused by electrical faults, which is particularly important in critical infrastructure applications such as data centers.

[0060] This embodiment further provides a charging method based on the aforementioned intelligent, compatible, and efficient charging control system. This method not only covers the basic charging process, such as vehicle identification, parameter setting, and intelligent algorithm invocation, but also incorporates two innovative steps: user behavior prediction and fault diagnosis. The specific implementation steps, in addition to basic connection and detection, include three stages: real-time monitoring of vehicle performance indicators, analysis of user usage habits, and proactive fault monitoring. Each stage relies on advanced data analysis technology and real-time communication networks, achieving a shift from passive response to proactive management, fundamentally improving charging efficiency and user satisfaction.

[0061] This embodiment provides a battery status monitoring system that includes a multimodal sensing module. A multimodal sensing module is a composite sensor array that integrates multiple physical quantity detection capabilities. Specifically, it can be implemented using a combination of optical fiber strain sensors and infrared thermal imagers to simultaneously capture the internal stress and surface temperature distribution of the battery. The multimodal sensing module is equipped with multiple sensors, including voltage sensors, current sensors, temperature sensors, and pressure sensors, to comprehensively monitor the battery's operating status. The multimodal sensing module collects real-time data on the battery's voltage, current, temperature, and expansion deformation. The voltage sensor detects the battery terminal voltage, the current sensor detects the battery's charge and discharge current, the temperature sensor detects the battery's surface and internal temperature, and the pressure sensor is installed at a specific location within the battery housing to monitor battery expansion deformation. The multimodal sensing module collects key battery data in real time. The corresponding smart charging device is connected to the multifunctional sensing unit included in the multimodal sensing module, which is responsible for collecting data such as the battery's internal voltage, current, ambient temperature, and expansion deformation. Specifically, the multifunctional sensing unit is equipped with a distributed optical fiber tension sensor to track changes in the battery stack's expansion rate. It also features a potentiometric impedance measurement instrument to periodically (≤30 seconds) capture the frequency response signal. Furthermore, the device features a contactless infrared sensor array to map the battery surface's real-time temperature.

[0062] Furthermore, the multimodal sensing module houses a data processing chip responsible for receiving the raw data transmitted by each sensor and performing preliminary processing. This processing includes, but is not limited to, filtering, digitization, and standardization to ensure data validity and accuracy. The processed data is transmitted in real time to a central monitoring unit, which determines the battery's operating status based on the received data. If an anomaly is detected, such as excessive temperature or a sudden voltage drop, protection mechanisms are immediately activated, such as automatically disconnecting the charging circuit or issuing an alarm signal.

[0063] The multimodal sensing module is also equipped with a wireless communication unit that supports wireless transmission methods such as Bluetooth and Wi-Fi. This allows battery status monitoring data to be uploaded to a cloud server in real time, facilitating remote monitoring and data analysis. Furthermore, to ensure long-term stable operation, the multimodal sensing module is designed with multiple levels of redundant protection, such as backup power supplies and failover mechanisms, to ensure that basic functionality can be maintained even if a component fails.

[0064] In this embodiment, the effectiveness of the multimodal sensing module is verified through experiments. Taking a lithium battery with a nominal capacity of 10Ah as an example, after 100 continuous charge and discharge cycles, the maximum temperature variation monitored by the multimodal sensing module is less than ±2°C, the maximum voltage deviation is less than ±0.1V, and the data transmission delay is less than 1ms. The above data all meet the industry standard requirements, verifying the reliability and high precision of this technical solution.

[0065] Combined with attachment Figure 2 This embodiment provides a battery health management method, which uses a multimodal sensing module to monitor the battery status, including the following steps: S1. Connect the multimodal sensing module to the battery and start monitoring; S2. Preprocess the collected data through a data processing chip to eliminate noise interference; S3. The central monitoring unit analyzes the preprocessed data to identify potential battery faults, such as overheating, overcharging, internal short circuit, etc., and records the time, type and severity of the fault; S4. For different fault types, the central monitoring unit initiates corresponding emergency response strategies, such as reducing the charge and discharge rate, starting the cooling system, disconnecting the circuit, etc.; S5. Regularly generate battery health reports, which include battery service life predictions, maintenance recommendations, etc., to help users take preventive measures in a timely manner and extend battery service life.

[0066] The above method further proposes to set up a battery maintenance module, establish a charging pile battery maintenance circuit and system, that is, in the charging pile circuit, an external resistor and capacitor can be used to form a Schmitt trigger or a monostable or multivibrator, that is, a core waveform generator is set up to achieve pulse waveform output, combine digital control and power electronics technology, and generate waveform voltage / current pulses according to the battery health report (see attached Figure 3 ), and proposes a maintenance strategy; using the waveform pulse generation mechanism of the multi-stage pulse repair method to intelligently control the charging waveform (positive pulse, negative pulse, and harmonic superposition) to achieve battery maintenance. The battery maintenance described here primarily utilizes a non-physical intelligent maintenance strategy. This multi-stage pulse repair method uses EIS (electrochemical impedance spectroscopy) to invert the charge transfer resistance and activate the enhanced harmonic mode. The negative pulse amplitude is dynamically adjusted based on the aging model.

[0067] At the same time, a set of intelligent charging process solutions was proposed, which specifically includes receiving information on the battery characteristics of newly connected vehicles and analyzing the open-circuit voltage relaxation phenomenon during discharge; using the support vector decomposition algorithm to separate the charging fluctuation components and identify battery aging; activating the transfer reinforcement learning function and designing a dedicated charging program with multiple linear constant temperature curves; using the gate-type cycle unit to predict the SOH degradation trajectory and adjust the charging stop point in a timely manner.

[0068] A further embodiment is proposed, featuring a networked system with intelligent, compatible, and efficient charging control systems that automatically adapt to different battery types, optimize charging efficiency, and extend battery life. This intelligent, networked charging system also incorporates a network collaboration component that utilizes a spatiotemporal convolutional grid model to comprehensively consider electricity price fluctuations and local loads. An improved multi-objective genetic algorithm is employed to determine the optimal charging speed ratio and grid balance point. When the system's operating efficiency is high, the electric vehicle grid supply mode is activated.

[0069] See attached Figure 4 This embodiment provides a cloud-based management system that connects multiple intelligent charging process solution systems via the internet, forming a charging network covering a wide area. Each charging station is equipped with one or more intelligent, compatible, and efficient charging control systems that meet the above-described requirements. The cloud server centrally manages data from each charging station, including charging records, device status, user feedback, and other information. It also provides a range of value-added services, including remote monitoring, fault alarms, and software upgrades.

[0070] Through this system, users can use a mobile app to view the location of nearby available charging stations, learn the number of available charging piles, schedule charging times, and even remotely control various charging parameters. This not only greatly facilitates users' ability to find and use charging piles, but also promotes the efficient allocation of charging resources. Based on the anonymous usage statistics of a large number of users on the cloud platform, researchers can also conduct more in-depth analysis and research to explore more intelligent and user-friendly charging solutions.

[0071] In summary, this invention provides an intelligent, compatible, and efficient charging control system and its application method. By integrating the latest information technology, automation technology, and artificial intelligence technology, it solves many problems existing in traditional charging methods, providing users with a more intelligent, convenient, and reliable option. With the rapid growth of the smart device and electric vehicle markets, this invention will demonstrate enormous market potential and social value in the future.

[0072] In response to the problems of poor compatibility and information silos in existing technologies, as well as the lack of a system charging strategy learning mechanism, the present invention further provides a cloud platform management system based on the above-mentioned intelligent compatible and efficient charging control system and method. The cloud-based policy management platform dynamically generates charging strategies applicable to various vehicle models. The cloud-based policy management platform is equipped with a vehicle type clustering engine that can create vehicle digital twins and achieve optimization based on the battery's electrochemical properties and charging graphs. The cloud-based policy management platform refers to a decision-making center deployed in a distributed server cluster and can be implemented using a microservices architecture. The charging strategy is generated by integrating the vehicle digital twin model with a reinforcement learning algorithm.

[0073] The cloud-based policy management platform also includes a reinforcement learning trainer that uses a multi-objective reward function encompassing charging speed improvement, temperature suppression, and battery life extension. The cloud-based policy management platform also features a transfer learning module that generates transfer learning policies, allowing improved policies for validated vehicle models to be replicated and applied to novel solid-state battery architectures.

[0074] The system integrates edge computing nodes, which are embedded systems within the charging station control unit or vehicle ECU. These nodes are implemented using automotive-grade chips with model compression technology to ensure real-time control in resource-constrained environments. These nodes, installed within the charging station or vehicle ECU, run a lightweight version of the charging control model system. These edge computing nodes run a compressed neural model suitable for TC397-class embedded system chips, enabling response times of less than 5 milliseconds. The nodes also include a voltage-temperature coordinated adjustment loop that dynamically adjusts the constant current / constant voltage transition point based on the battery's remaining capacity. Furthermore, a built-in gradient penalty mechanism ensures that the reinforcement learning approach employed does not deviate from the battery's permitted operating range. When deployed within the charging station or vehicle ECU (Electronic Control Unit), the edge computing node executes the lightweight charging control model. This embedded lightweight charging control model analyzes the current battery status and transmits processed charging instructions via the CAN bus to the central processing module to guide the formulation of the charging control strategy. In-vehicle edge computing nodes leverage local computing power to reduce latency, enabling rapid response even in emergency situations and ensuring safe and efficient charging. For example, if the vehicle battery temperature exceeds a preset threshold, the edge computing node immediately notifies the central processing module, which then adjusts the charging status, potentially reducing charging power or suspending charging to avoid potential safety risks.

[0075] In vehicle-based scenarios, edge computing nodes are deployed within the vehicle's ECU and can directly read vehicle battery status parameters such as SOC (State of Charge) and SOH (State of Health). These nodes process charging data from the electric vehicle in real time, optimizing the charging process based on this data for rapid response and intelligent management. Local processing by edge computing nodes reduces the computing burden on the cloud, improving system responsiveness and efficiency. The specific deployment location of edge computing nodes can be selected within the charging station or within the vehicle's ECU to meet different scenario requirements.

[0076] When used at charging stations, edge computing nodes are deployed inside the charging stations. Their function is to monitor and control the various electric vehicles connected to the charging stations in real time. Through high-speed data exchange, they work in conjunction with the charging station's central processing module to ensure a smooth charging process. Furthermore, edge computing nodes can dynamically adjust charging strategies based on the actual power supply conditions in the area where the charging stations are located, improving energy utilization. For example, they can reduce charging power during peak grid load periods and increase it during off-peak periods, effectively smoothing out power peaks and contributing to the stable operation of the power system.

[0077] Specifically, the lightweight charging control model of the edge computing node is developed based on a machine learning algorithm and can predict the optimal charging strategy, including charging timing and charging rate, based on historical data. The edge computing node collects data from each charging process and continuously optimizes the model, improving charging efficiency while reducing costs. For example, by analyzing the battery's historical charging records and current charging demand, the lightweight charging control model can automatically identify the optimal charging time window to avoid the impact of high electricity prices during peak hours. In addition, the edge computing node also supports remote monitoring, allowing operators to view charging status at any time through the cloud platform, conduct troubleshooting and maintenance scheduling, and enhance the manageability and reliability of the system.

[0078] Furthermore, the cloud-based policy management platform in this embodiment primarily includes a data acquisition module, an analysis and processing module, and a policy generation module. The data acquisition module is responsible for acquiring real-time vehicle information and user demand information from multiple vehicle management systems, user feedback channels, and vehicle terminals. The analysis and processing module comprehensively analyzes and processes the received information to determine whether the vehicle's current status meets specific charging requirements, such as remaining battery capacity, health status, and temperature. The policy generation module dynamically generates personalized charging policies for different vehicle types based on the analysis and processing results. Specifically, when a vehicle connects to a charging station, the system first automatically identifies the vehicle model and battery type through wireless communication technology. The cloud-based policy management platform then intelligently analyzes and rapidly generates an optimized charging plan based on stored user charging history and maintenance information, combined with the vehicle's specific conditions (including but not limited to environmental factors and vehicle load). For example, for electric vehicle A, which uses a high-energy-density lithium battery pack, the policy generation module might recommend a fast charging mode to meet the user's time efficiency needs. For electric vehicle B, which uses a standard-capacity lead-acid battery, a gentle charging mode is recommended to prolong battery life. In addition, the strategy generation process fully considers factors such as grid load and peak-valley electricity price differences, striving to reduce costs while meeting user charging needs. In this embodiment, the cloud-based strategy management platform is not limited to generating charging strategies but also provides remote monitoring and fault diagnosis functions, allowing users to view charging status, historical data, and receive abnormal alarm information at any time, improving user experience and system reliability. Overall, by introducing cloud computing technology and big data analysis methods, the compatible charging solution provided by this invention can achieve more flexible, efficient, and intelligent vehicle charging management, greatly promoting the development of the new energy vehicle industry.

[0079] The cloud-based policy management platform also incorporates a federated learning framework, a distributed machine learning paradigm that protects data privacy. This framework utilizes differential privacy encryption and model parameter aggregation to facilitate cross-institutional data collaboration. Specifically, when a new electric vehicle is connected to a charging station, a multimodal sensing module immediately initiates full-dimensional data collection. Fiber optic strain sensors monitor the battery module's expansion rate, and an infrared array generates a temperature distribution thermogram. This data is pre-processed by edge nodes and uploaded to the cloud.

[0080] The cloud-based policy management platform leverages the cross-automaker knowledge base stored in the federated learning framework to match the current battery's relaxation voltage characteristics with historical optimization strategies. After generating a charging plan with a dynamic constant current threshold, it is converted into executable instructions by edge computing nodes. During the charging process, the edge node continuously monitors the battery status. When an abnormal temperature rise is detected, the local control loop is immediately triggered to adjust the current output. The operating condition data is encrypted and uploaded to the federated model for updating. This facilitates data connectivity between automakers, energy suppliers, and charging station operators. Within this federated learning framework, differential privacy protection measures are implemented to enhance data security. A fuzzy inference fusion layer is implemented to integrate power consumption forecasts and support patterns. Scaling update rules are also developed to adjust the charging intensity benchmark based on fluctuations in the electricity market. Real-time data sharing enables real-time adjustment of personalized charging plans, effectively improving charging efficiency and ensuring charging safety. It also promotes data interaction between power grids, addressing the existing charging system's poor adaptability to different vehicle models and insufficient data collaboration capabilities, providing a more intelligent and secure charging solution for new energy vehicles.

[0081] Compared with existing technologies, traditional charging systems rely on manual experience to set charging curves and cannot automatically adapt to new battery systems. Existing stand-alone BMS can only process local sensor data and lack cross-vehicle data collaboration capabilities. Most charging piles use fixed-parameter PID control, which makes it difficult to cope with parameter drift caused by battery aging. This solution builds a hierarchical decision-making system to achieve global strategy optimization in the cloud, ensure real-time control at the edge, break down data barriers through federal mechanisms, and form an adaptive and evolving charging management closed loop. Through the above technical solutions, this application realizes safe and fast charging of electric vehicles of multiple brands, shortening charging time while ensuring the health of the battery.

[0082] When connected to a new solid-state battery vehicle, the system automatically utilizes transfer learning strategies to generate an adaptive charging curve, eliminating downtime and waiting due to manual commissioning. During peak grid load periods, the system dynamically adjusts the charging power baseline to balance charging demand and grid stability. For battery packs with localized aging, the system identifies areas of abnormal temperature rise and automatically reduces the charging current of the corresponding modules, effectively preventing the risk of thermal runaway.

[0083] Furthermore, the cloud-based policy management platform proposed in this embodiment also includes a vehicle clustering engine, a reinforcement learning trainer, and a transfer learning module. The vehicle clustering engine is a component that classifies vehicles based on battery electrochemical characteristics and charging curves and creates a digital twin model. This is achieved by combining a clustering algorithm with analysis of battery open-circuit voltage relaxation characteristics, providing a classification basis for generating charging strategies.

[0084] A reinforcement learning trainer refers to a strategy generation component that takes charging efficiency, temperature rise suppression, and life extension as optimization objectives. Specifically, it can be implemented using a multi-objective reward function with dynamic weight adjustment to balance the constraints between different optimization objectives. Based on the multi-objective reward function, the reinforcement learning trainer dynamically allocates weights in the three dimensions of charging efficiency, temperature rise suppression, and life extension to generate a charging strategy that meets multi-dimensional constraints. Further, based on the compression model of the neural architecture search, the response delay is maintained in the automotive-grade chip, and the voltage and temperature are jointly controlled in the loop. The constant current and constant voltage thresholds in the fast charging stage are dynamically adjusted according to the real-time state of charge. While adopting the multi-objective reward function with dynamic weight adjustment, the gradient penalty mechanism is also used in this embodiment to constrain the reinforcement learning strategy within the battery safety operation boundary.

[0085] The gradient penalty mechanism imposes constraints on the policy gradient during reinforcement learning policy training. Specifically, Lagrange multipliers can be used to impose hard constraints on the thermal stability limits of battery materials. This feature prevents charging strategies from exceeding the physical tolerances of the battery, ensuring operational safety. Neural architecture search uses automated machine learning algorithms to search for the optimal neural network structure. Specifically, multi-objective optimization algorithms can be used to balance model accuracy and computational complexity to achieve model compression. Building on the reinforcement learning algorithm, neural architecture search forms a closed loop from policy generation to hardware deployment. The reinforcement learning trainer trains the charging policy using a multi-objective reward function, while neural architecture search compresses the policy network for efficient deployment. Reinforcement learning is the policy's "intelligent engine," responsible for generating multi-objective optimized charging control logic within safety boundaries. Neural architecture search is the policy's "slimming tool," transforming complex policies into deployable, lightweight models. These two components form a complete chain from algorithm design to engineering implementation: reinforcement learning provides intelligent decision-making capabilities, while neural architecture search ensures real-time execution in resource-constrained vehicle environments. Together, they achieve "safe, efficient, and long-lasting" battery fast charging control.

[0086] This feature is used to reduce the complexity of the algorithm and ensure that the real-time requirements are met on automotive-grade chips. Among them, the voltage-temperature joint control loop refers to a feedback system that uses battery voltage and temperature parameters as collaborative control variables. Specifically, a proportional integral differential controller can be used in combination with a thermodynamic model to establish a dynamic adjustment mechanism. This feature is used to adjust the charging parameters according to the real-time status of the battery to adapt to nonlinear changes under different working conditions. The transfer learning module refers to migrating the optimized strategy parameters of a verified vehicle model to the components of a new battery system. Specifically, it can be achieved by using a feature space mapping algorithm combined with solid-state battery electrochemical characteristic adaptation to shorten the development cycle of new battery charging strategies. The transfer learning module analyzes the mapping relationship between solid-state batteries and existing lithium-ion batteries in the electrochemical feature space, and migrates the key parameters of the verified strategy to the new battery system. The adaptation can be completed by only a limited number of online calibrations of the migrated strategy.

[0087] Compared to existing technologies, traditional charging strategy generation methods typically employ fixed-weight optimization or single-objective control, failing to dynamically balance conflicting relationships among multiple objectives. Existing strategy migration techniques rely on the physical similarity of battery systems, making them difficult to adapt to novel electrochemical systems such as solid-state batteries. This solution addresses the multi-dimensional optimization conflict issue through a multi-objective dynamic weight adjustment mechanism, utilizing feature space mapping to overcome the strategy migration barriers caused by differences in battery systems.

[0088] Through the above technical solution, this application realizes the dynamic adaptation of battery characteristics of multiple models, solves the problem of rapid migration of charging strategies of different electrochemical systems, and optimizes the balance between charging efficiency and battery life while ensuring charging safety.

[0089] Specifically, a compression model optimized through neural architecture search is deployed on automotive-grade chips, and the control instruction generation delay is controlled to the millisecond level through lightweight algorithm processing, meeting the real-time response requirements of fast charging scenarios. The essence of the lightweight charging control model is to reduce system complexity and resource consumption while ensuring charging efficiency through algorithm optimization and architecture reconstruction. Its core features are reflected in the following: Dynamic power adaptation mechanism: By real-time monitoring of battery status (such as SOC, SOH) and external environmental parameters, fuzzy logic control1 or improved PID algorithm14 is used to achieve dynamic adjustment of charging parameters. Hierarchical control architecture: The use of a hierarchical control architecture increases the system modularity by 30% and the hardware resource reuse rate to 85%; Protocol compatibility optimization: The use of a variant firefly algorithm reduces the code volume of the protocol parsing module by 50% while maintaining multi-protocol support capabilities.

[0090] A joint control loop for voltage and temperature parameters collects the battery state of charge in real time and dynamically adjusts the current and voltage thresholds (e.g., 800V) during the fast-charging phase based on preset battery characteristic curves. For example, it automatically reduces the peak current during the constant-current phase in low-temperature environments. A gradient penalty mechanism imposes constraints during the reinforcement learning policy update process, converting physical parameters such as the battery material's expansion coefficient and the electrolyte decomposition temperature into mathematical constraints to ensure that the generated charging strategy remains within a safe operating range. Compared to existing technologies, traditional edge computing nodes use fixed-threshold charging control strategies that are unable to adapt to dynamic factors such as battery aging and temperature fluctuations, and lack safety constraints for policy learning. This solution, through a dual mechanism of dynamically adjusting thresholds and safety constraints, improves control accuracy and safety while maintaining real-time responsiveness. Existing reinforcement learning strategies are prone to dangerous operations that exceed physical boundaries. This solution, through a gradient penalty mechanism, converts material properties into mathematical constraints, fundamentally eliminating safety hazards. Through the above technical solution, this embodiment realizes charging control with millisecond-level response on the edge computing node, and can dynamically adjust the charging parameters according to the real-time status of the battery. At the same time, mathematical constraints are used to ensure that all control strategies meet the safe operation requirements of battery materials, solving the technical problems of excessive delay, poor adaptability and safety hazards in traditional methods.

[0091] This embodiment also proposes a technical solution for a multimodal sensing module integrating a distributed fiber optic strain sensor, an electrochemical impedance spectroscopy acquisition unit, and a non-contact infrared array. The distributed fiber optic strain sensor is a device that captures microscopic deformation rates through a fiber optic network covering the surface of the battery module. Specifically, it can be implemented using a Bragg grating sensor array. The strain distribution is calculated by measuring the wavelength shift of the optical signal, which is used to determine the mechanical stability of the battery structure. The electrochemical impedance spectroscopy acquisition unit is a device that collects frequency domain response data at a fixed frequency. Specifically, it can be implemented using a combination of an AC excitation signal generator and a lock-in amplifier. The charge transfer state within the battery is analyzed by analyzing impedance phase changes at different frequencies. The non-contact infrared array is a sensor that generates a temperature field distribution through thermal radiation detection. Specifically, it can be implemented using an array of microbolometers. It generates a two-dimensional temperature distribution map by receiving infrared radiation intensity at different locations on the battery surface. The distributed fiber optic strain sensor is placed at the junction of the battery module housing and the tab. When lithium dendrites grow or the electrode expands, the fiber optic network generates a wavelength shift signal related to the deformation. This signal is demodulated and converted into strain rate data. During the charging process, the electrochemical impedance spectroscopy acquisition unit periodically injects a multi-band AC signal. By measuring the phase difference between voltage and current, it calculates the charge transfer resistance and diffusion impedance, thereby identifying electrolyte decomposition or SEI thickening. A non-contact infrared array, mounted above the battery pack at a top-down angle, scans the infrared radiation intensity emitted from each cell surface, generating a distribution map that includes temperature gradients and abnormal hotspots. Data from these three sensor types is synchronized with timestamps and fed into edge computing nodes to form a three-dimensional state matrix of mechanical deformation, electrochemical impedance, and temperature distribution. Compared to existing technologies, traditional solutions typically rely solely on voltage, current, and a small number of point-type temperature sensors for monitoring, failing to capture changes in mechanical stress and abnormal temperature field distribution within the battery. This solution utilizes a fiber optic sensor network for global strain monitoring, combined with frequency-domain impedance analysis to reveal detailed electrochemical processes. Combined with array-based infrared temperature measurement, it eliminates monitoring blind spots, providing 15-minute advance warning of thermal runaway risks and accurately identifying charging compatibility issues caused by battery aging. Through the above technical solution, this application realizes full-dimensional real-time monitoring of the mechanical deformation, electrochemical properties and temperature distribution of the battery pack, effectively identifying structural damage caused by electrode expansion, impedance anomalies caused by electrolyte decomposition, and thermal runaway risks caused by local overheating.

[0092] Furthermore, the vehicle clustering engine mentioned above collects the electrochemical characteristics of vehicle batteries and charging curves to build a vehicle digital twin model. This model accurately reflects the performance differences between different vehicle models during the charging process. Specifically, the vehicle clustering engine includes not only the data acquisition module and data analysis module in the cloud-based policy management platform, but also a digital twin generation module for processing and generation. The data acquisition module collects vehicle battery status data, such as key parameters such as battery temperature, charging current, and battery voltage. The data analysis module analyzes the collected data, applies a clustering algorithm to group vehicles with similar characteristics into the same category, and establishes corresponding charging strategies based on the category characteristics. The digital twin generation module generates a digital twin model of the vehicle based on the analysis results, facilitating real-time monitoring and optimization of the charging process. In this embodiment, the data acquisition module acquires battery data through an onboard sensor network and transmits it to the data analysis module via wireless or wired means. The onboard sensor network includes not only sensors in the multimodal sensing module but also sensors in the intelligent charging device. The data analysis module uses a K-means clustering algorithm to classify vehicles, with each category corresponding to a different charging strategy, making the charging process more personalized and efficient.

[0093] Furthermore, the data analysis module analyzes the vehicle battery charging curve to identify characteristic parameters of different charging stages, such as charging efficiency and maximum charging power, to optimize the charging strategy. The charging curve feature extraction methods used in this embodiment include but are not limited to mathematical tools such as Fourier transform and wavelet transform. These methods identify the charging mode of different vehicle models by extracting key feature points of the charging curve. The specific algorithm can be expressed as:

[0094]

[0095] in, represents the characteristic vector of the charging curve, For the The values ​​of the characteristic points are used to effectively distinguish the charging characteristics of different vehicle types, and then customize the optimal charging plan for each category. In experimental tests, the clustering method proposed in this embodiment can correctly classify vehicles with an accuracy rate of over 95%, ensuring the effectiveness and targetedness of the charging strategy.

[0096] The system collects data through the on-board sensor network (temperature / voltage / current and other sensors) and the intelligent charging device. The multimodal sensor fusion technology can control the data loss rate to below 0.3% and supports the concurrent access of 100,000 devices.

[0097] A composite feature extraction method using wavelet transform and Fourier transform improves curve feature recognition by 15% compared to traditional methods. The K-means algorithm's classification accuracy drops to 82% when the battery temperature exceeds 45°C, requiring compensatory optimization using hierarchical clustering. A fuzzy control algorithm based on real-time SOC (State of Charge) is designed to control the fuel cell system, ensuring stable operation and producing the same unit power with less hydrogen. A dynamic current adjustment strategy is implemented, shortening charging time by 23% in high-temperature environments and keeping the temperature rise within Δ8°C.

[0098] Using multi-physics coupling modeling, the model integrates electrochemical-thermodynamic coupling equations:

[0099]

[0100] Through real-time parameter calibration, the voltage prediction error is <0.05V.

[0101] By designing a charging control device that automatically identifies and is compatible with multiple charging standards, the existing technology overcomes the increased cost and system complexity associated with adding conversion modules, achieving more efficient, stable, and manageable charging control. Test data shows that this mechanism reduces policy adjustment response speed to seconds (1.2 seconds on average). The personalized settings module supports presets for various charging scenarios.

[0102] Local edge nodes process 80% of real-time control commands, reducing cloud communication load by 62%. However, this design aims to improve system reliability and stability. Using a dual verification mechanism combining adaptive Kalman filtering and blockchain, the accuracy of abnormal data identification reaches 99.2%. Experimental data shows that this technology can save approximately 12,000 yuan in annual electricity bills per charging pile (based on a daily average of 50 kWh).

[0103] The vehicle digital twin model in this embodiment not only contains static data but also supports dynamic adjustment. By monitoring the status of the vehicle battery in real time, the digital twin model is updated to cope with the impact of changes in vehicle usage conditions. For example, when the vehicle is in a high-temperature environment for a long time, the model automatically adjusts the charging strategy to reduce high-power charging time and avoid battery overheating and damage. In addition, the vehicle digital twin model also supports user-defined settings, allowing users to adjust charging priority and charging speed according to their personal needs.

[0104] Furthermore, the system provided in this embodiment also includes a user interface that displays the vehicle's current charging status, recommended charging strategies, and the optimal charging plan for the next period of time. This interface graphically presents the vehicle's charging curve, allowing users to intuitively understand the charging process and select the recommended strategy. Specifically, users can select fast charging mode or economic charging mode in the interface, and the system will automatically adjust the charging curve based on the selection to achieve the optimal charging effect.

[0105] Traditional charging methods typically employ fixed-weight optimization or single-objective control, failing to dynamically balance conflicting relationships among multiple objectives. This makes them difficult to adapt to new electrochemical systems like solid-state batteries. Our invention addresses this multi-dimensional optimization conflict through a multi-objective dynamic weight adjustment mechanism.

[0106] Most charging piles in the existing technology use fixed parameter PID control, which makes it difficult to cope with parameter drift caused by battery aging. Our invention realizes global strategy optimization in the cloud by constructing a multi-level decision-making system, ensures real-time control at the edge, breaks down data barriers through a federal mechanism, and forms an adaptive and evolving charging management closed loop. This application realizes safe and fast charging of electric vehicles of multiple brands, shortening the charging time while ensuring the health of the battery. The solution of the present invention dynamically generates charging strategies for multiple models through a cloud-based strategy platform, real-time control of edge computing nodes, multi-modal sensor data collection, and a federated learning framework to connect data islands. It solves the problems of poor compatibility and insufficient data collaboration of traditional charging systems, improves charging efficiency, ensures battery safety, and promotes grid interaction, meets the specific electrochemical requirements of batteries of different models, realizes personalized adjustment of charging strategies, and at the same time strengthens data interoperability between information islands, promoting the development of charging technology in a smarter and safer direction.

[0107] In summary, this invention provides an intelligent, compatible, and efficient charging control system and its application method, which also constitutes a cloud platform management system. By integrating the latest information technology, automation technology, and artificial intelligence technology, it solves many problems existing in traditional charging methods, providing users with a more intelligent, convenient, and reliable option. With the rapid growth of the smart device and electric vehicle markets, this invention will demonstrate enormous market potential and social value in the future.

Claims

1. Intelligent compatible and efficient charging control system, characterized by: include: Connecting seat, used for connecting with external tram; A central processing module, connected to the connection socket, for receiving and processing information including the charging power of the electric vehicle; A locking module is provided between the central processing module and the connection socket, and is used to lock the connection after power matching; An adaptation module, including a voltage regulation unit and a preset unit, for adjusting the voltage to match the tram requirements; A control module, connected to the central processing module and the locking module, for adjusting the locking state; The control module consists of a CAN bus, a voltage stabilizing unit and a control unit, wherein the CAN bus is connected to the voltage stabilizing unit and the control unit respectively, and its signal output end is connected to the locking module, and the control unit is connected to the central processing module. The logic control module is located between the connector and the central processing module and is used to coordinate data interaction; The intelligent algorithm module includes a core computing unit and a data storage unit. The intelligent algorithm module is connected to the central processing module and executes the instructions of the central processing module. It is used to analyze the historical usage data of the electric vehicle and predict the future power consumption pattern, thereby intelligently adjusting the charging strategy. At the same time, a battery maintenance module is also provided, and the battery charging maintenance adopts a non-physical intelligent maintenance strategy and a multi-stage pulse repair method.

2. The intelligent compatible and efficient charging system according to claim 1, characterized in that: A multimodal sensing module is also provided, which is equipped with multiple sensors, including a voltage sensor, a current sensor, a temperature sensor, and a pressure sensor, for comprehensively monitoring the working status of the battery; It further includes: a fiber optic tension sensor for monitoring the expansion deformation rate of the battery module; an electrochemical impedance spectroscopy acquisition unit for collecting frequency domain response signals at a time interval not exceeding 30 seconds; a non-contact infrared sensor array for real-time mapping of the temperature distribution on the battery surface; and the multimodal sensing module for real-time acquisition of battery voltage, current, temperature and expansion deformation data.

3. The charging method based on the above intelligent compatible and efficient charging control system is characterized in that: Including connecting the tram to the connecting seat; The central processing module detects the power of the tram and compares it with the preset parameters. Based on the comparison results, it decides whether to lock the tram directly or adjust the voltage first and then lock it. Repeat the detection and adjustment until the electric vehicle can complete the entire charging process at constant power. When the electric vehicle is initially connected to the connection socket, the central processing module sets the parameters of the preset unit according to the detected electric vehicle power; when it is detected that the connection socket power matches the preset parameters of the electric vehicle, the central processing module drives the locking module through the control module to achieve locking and start charging; When it is detected that the power of the connection socket does not match the preset parameters of the tram, the central processing module also instructs the boost module in the adapter module to adjust the voltage through the control module until it meets the preset parameters, and then drives the locking module to execute the locking action again to start charging. If the docking station power does not match the tram's preset parameters, the system will issue a warning through the display unit.

4. The charging method based on the above intelligent compatible and efficient charging control system is characterized in that: It also includes a battery health management method that uses a multimodal sensing module to monitor battery status. The method comprises the following steps: S1, connecting the multimodal sensing module to the battery and starting monitoring; S2. Pre-process the collected data through the data processing chip to eliminate noise interference; S3. The central monitoring unit analyzes the pre-processed data to identify potential battery failures, such as overheating, overcharging, and internal short circuits, and records the time, type, and severity of the failure. S4. For different fault types, the central monitoring unit initiates corresponding emergency response strategies, such as reducing the charge and discharge rate, starting the cooling system, disconnecting the circuit, etc. S5. Generate battery health reports regularly, including battery life predictions, maintenance recommendations, etc., to help users take preventive measures in a timely manner and extend battery life.

5. The charging method based on the intelligent compatible and efficient charging control system according to claim 4 is characterized in that: It also includes battery maintenance methods; a core waveform generator is set in the circuit of the charging pile to provide a maintenance strategy based on the battery health report; and the charging waveform is intelligently controlled through the waveform pulse generation mechanism in the multi-stage pulse repair method to achieve battery maintenance.

6. Cloud platform management system based on intelligent compatibility and efficient charging control system, characterized by: include: The cloud-based policy management platform includes a data acquisition module, an analysis and processing module, and a policy generation module. The data acquisition module is responsible for obtaining vehicle information and user demand information in real time from multiple vehicle management systems, user feedback channels and vehicle terminals; The analysis and processing module performs comprehensive analysis and processing on the received information to determine whether the current state of the vehicle meets specific charging requirements, such as the remaining battery capacity, health status, and temperature. The strategy generation module dynamically generates personalized charging strategies for different vehicle types based on the analysis and processing results; The cloud-based policy management platform further includes: A vehicle clustering engine creates a digital twin of the vehicle and optimizes charging curve characteristics based on the battery's electrochemical properties and charging graphs. The reinforcement learning trainer generates a strategy component that optimizes charging efficiency, temperature rise suppression, and life extension. This strategy is implemented using a multi-objective reward function with dynamic weight adjustment. A gradient penalty mechanism is used to balance the constraints between different optimization objectives, ensuring that the reinforcement learning strategy remains within the battery's safe operating range. Transfer learning module, which enables the migration of optimized strategy parameters of verified vehicle models to new battery systems; Edge computing nodes, deployed in charging piles or vehicle-mounted ECUs, are used to execute lightweight charging control models; A multimodal sensing module, which is used to collect data on battery voltage, current, temperature, and expansion deformation in real time; A federated learning framework is used to connect data from automakers, power grid operators, and charging facilities.

7. The cloud platform management system according to claim 6, characterized in that: The reinforcement learning trainer performs dynamic weight allocation based on a multi-objective reward function to generate a charging strategy that meets multi-dimensional constraints. It further maintains response delay and voltage and temperature joint control loops in automotive-grade chips based on the compression model of neural architecture search, and dynamically adjusts the constant current and constant voltage thresholds in the fast charging stage according to the real-time state of charge.

8. The cloud platform management system according to claim 6, characterized in that: The multimodal sensing module integrates the technical solutions of distributed optical fiber strain sensors, electrochemical impedance spectroscopy acquisition units and non-contact infrared arrays. The data of the three types of sensors are synchronized through timestamps and input into the edge computing node to form a three-dimensional state matrix of mechanical deformation, electrochemical impedance and temperature distribution.

9. The cloud platform management system according to claim 6, characterized in that: The vehicle digital twin model not only contains static data but also supports dynamic adjustment.

Images