Intelligent control system and method for mobile energy storage charging robot

Abstract

The invention provides an intelligent control system and method for a mobile energy storage charging robot, and the system employs a monitoring core board as a control center, and integrates hardware resources of an energy storage system, a power system, a chassis driving module, a BMS, a communication module, a sensor group, and a man-machine interaction module. And software functions of data acquisition, preprocessing, feature extraction, decision control, cloud interaction and operation and maintenance optimization are combined to realize full-process intelligent control of the mobile energy storage charging robot. According to the method, a robot is used as an execution main body, computer data processing logic of data acquisition, data processing, feature extraction, decision judgment, execution control and feedback optimization is followed, and multi-source data fusion, intelligent algorithm modeling and dynamic decision control are carried out. Charging compatible adaptation, accurate battery management, intelligent scheduling, efficient charging and cloud collaborative operation and maintenance are realized.

Description

Technical Field

[0001] This invention relates to the field of mobile energy storage and charging technology, and in particular to an intelligent control system and method for mobile energy storage and charging robots. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the problem of lagging construction of charging infrastructure has become increasingly prominent. Traditional fixed charging piles have shortcomings such as limited layout, uneven utilization, and high expansion costs. Mobile energy storage charging robots, as a flexible and convenient energy replenishment device, can realize the transformation from vehicle-to-charging-pile to charging-pile-to-vehicle, becoming an important solution to the charging difficulty problem.

[0003] Currently, several mobile energy storage and charging robot products have emerged both domestically and internationally, but their technology still has many shortcomings, mainly in the following aspects: Current mobile energy storage and charging robots are mostly scheduled using manual assignment or simple distance-first strategies, without considering dynamic scheduling based on multi-dimensional information such as grid peak and off-peak times, their own energy storage status, vehicle charging demand priority, and environmental factors. For example, during grid peak hours, robots still discharge at high power, exacerbating grid load; when multiple robots work together, uneven resource allocation can easily occur, with some robots idle while some vehicles wait for too long; path planning only considers straight-line distance, without fully taking into account factors such as slope, obstacles, and energy consumption, resulting in low driving efficiency and high energy consumption.

[0004] Meanwhile, the operation and maintenance of existing mobile energy storage and charging robots mainly rely on manual inspections, lacking data-driven proactive early warning and remote operation and maintenance capabilities. Equipment failures (such as power module failures, chassis motor failures, battery insulation failures, etc.) are often only discovered after they occur, resulting in long downtime and impacting operational efficiency; the lack of long-term trend analysis of equipment status makes it impossible to predict potential failures, leading to high operation and maintenance costs.

[0005] Secondly, existing systems only collect basic data such as battery voltage and current, lacking comprehensive data collection on multiple dimensions such as vehicle charging needs, environmental parameters, and equipment status, resulting in insufficient decision-making basis. For example, information such as the battery type and charging protocol version of the vehicle to be charged is not collected, making dynamic protocol adaptation impossible; parameters such as ambient temperature and altitude are not collected, causing charging strategies and route planning to be unable to adapt to environmental changes. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the defects of the existing technology. The present invention proposes an intelligent control system and method for mobile energy storage and charging robots.

[0007] To address the aforementioned issues, existing mobile energy storage charging robots primarily rely on manual assignment or simple distance-first strategies for scheduling, failing to incorporate dynamic scheduling based on multi-dimensional information such as grid peak / valley times, their own energy storage status, vehicle charging demand priority, and environmental factors. Furthermore, the operation and maintenance of existing mobile energy storage charging robots mainly depend on manual inspections, lacking data-driven proactive early warning and remote maintenance capabilities. Equipment failures are often only discovered after they occur, leading to long downtimes and impacting operational efficiency. The lack of long-term trend analysis of equipment status makes it impossible to predict potential failures, resulting in high maintenance costs. Secondly, existing systems only collect basic data such as battery voltage and current, lacking comprehensive data collection on multi-dimensional factors such as vehicle charging demand, environmental parameters, and equipment status, leading to insufficient decision-making basis. Therefore, the technical solution adopted in this invention is: An intelligent control system for a mobile energy storage and charging robot, with the mobile energy storage and charging robot as the executing entity, includes the following steps: Data acquisition steps: The robot collects multi-source initial data in real time through its onboard sensor group, communication module and interaction interface. The multi-source initial data includes the robot's own status data, the charging demand data of the electric vehicle to be charged, external environment data and scheduling instruction data issued by the cloud platform. Data preprocessing steps: The robot performs noise reduction, standardization, anomaly detection, and data completion on the multi-source initial data to obtain standardized data; Feature extraction step: Based on the standardized data, the robot extracts multi-dimensional feature data through a preset feature extraction algorithm. The multi-dimensional feature data includes battery status features, charging demand adaptation features, environmental adaptation features, and location scheduling features. Decision-making steps: The robot inputs the multi-dimensional feature data into the preset decision model, and performs charging compatibility judgment, battery safety status judgment, charging strategy matching judgment, path planning judgment and operation and maintenance early warning judgment respectively, and outputs the corresponding decision instructions; Execution control steps: The robot executes corresponding control operations according to the decision instructions, including movement control, dynamic adjustment of charging parameters, battery protection control, cloud interaction control, and human-machine interaction control; Feedback optimization steps: The robot collects the execution feedback data of the control operation, and combines it with the optimization instructions of the cloud platform to iteratively optimize the feature extraction algorithm, decision model and control parameters, and update the preset algorithm and model parameters.

[0008] Preferably, the robot's own state data includes: Energy storage system data: individual cell voltage, total voltage, charge / discharge current, individual cell temperature, total temperature, initial SOC, initial SOH, battery pack insulation resistance, coolant flow rate and temperature; Power system data: power module input voltage, output voltage, output current, module temperature, operating efficiency, and cooling fan speed; Chassis drive module data: rated power, peak power, rated torque, peak torque, speed, current, voltage, chassis attitude angle, driving speed, driving distance, steering angle, braking status, obstacle crossing status of chassis motor; Equipment status data: monitoring the operating status of the core board, the signal strength and connection status of the communication module, the operation records of the human-machine interaction module, and payment verification data.

[0009] Preferably, the vehicle charging demand data includes: the battery type, rated battery capacity, voltage range, charging power demand, initial SOC, target SOC, charging protocol type, vehicle location coordinates, and vehicle identification information of the vehicle to be charged. External environmental data includes: ambient temperature, ambient humidity, altitude, atmospheric pressure, precipitation status, and obstacle information; the scheduling instruction data issued by the cloud platform includes: scheduling priority, peak and valley time period identifiers, charging price information, operation and maintenance task instructions, and algorithm update instructions.

[0010] Preferably, the data preprocessing step specifically includes: Noise Reduction: The Kalman filter algorithm is used to reduce noise in the battery cell voltage, current, and temperature data. The filtering equation is: Prediction equation: ; Filter gain equation: ; Update equation: ; in, The data at time k is the filtered data. Here is the state transition matrix. For the input matrix, Input data for time k-1. For filter gain, To predict the error covariance, For the observation matrix, To observe the noise covariance, The data is the observation data at time k; Standardization: Min-max standardization is used to normalize data such as voltage range and power demand. The formula is: ; in, For standardized data, The original data, The minimum value of the data. The maximum value of the data; Anomaly detection: The 3σ criterion is used to remove outliers from the data. When the data meets the criteria... When it is determined to be an outlier, among which The mean of the data. The standard deviation of the data; Data completion: Missing data is completed using linear interpolation. The formula is as follows: ; in, For missing data, , These are the valid data before and after the missing data, respectively. , , These represent the times when the corresponding data was collected.

[0011] Preferably, the feature extraction step specifically includes: Battery state feature extraction: SOC characteristics: SOC is estimated based on the ampere-hour integral method combined with the extended Kalman filter (EKF), and the formula is as follows: ; in, For the initial SOC, For the battery's rated capacity, This is the charging and discharging current. For EKF correction terms; SOH characteristics: SOH is estimated based on a nonlinear regression model, using the following formula: ; in, For the current battery capacity, This refers to the battery's internal resistance. The number of loops. This represents the average battery temperature. ~ , For model parameters, This is the error term; Battery consistency characteristics: Calculating the standard deviation of individual cell voltages and temperature standard deviation Consistency coefficient ,in This refers to the battery's nominal voltage. Charging demand adaptation feature extraction: compatibility ; in, For voltage range matching, For power demand matching, For charging protocol compatibility, , , These are the weighting coefficients. ; Environmental adaptation feature extraction: Environmental adaptability coefficient ; in, For temperature adaptability coefficient, Humidity adaptation factor, This refers to the altitude adaptation factor. , , These are the weighting coefficients; Location scheduling feature extraction: Path length feature: Calculate the straight-line distance based on the current location and the coordinates of the target parking space. ; Path cost characteristics: Calculate path cost by considering slope and obstacle crossing conditions. ,in For distance, The slope coefficient is... The obstacle crossing coefficient, For energy consumption, , , , These are the weighting coefficients.

[0012] Preferably, the decision-making step specifically includes: Charging compatibility assessment: When the compatibility is... When it is determined to be compatible, a "Charging Allowed" command is output; when When it is determined to be partially compatible, a "reduce power charging" command is output; when If the condition is not met, a "refuse charging" command will be output. Battery safety status assessment: Overvoltage detection: When the voltage of a single battery cell... When this occurs, an overvoltage protection command is output; Over-temperature judgment: When the battery temperature When, output the liquid cooling start command; when When the time comes, output a stop charging command; Insulation judgment: When the insulation resistance When this occurs, output the "insulation fault" command; Charging strategy matching judgment: Constant current mode: when and At that time, a constant current charging command is output, and the charging current is... ; Constant pressure mode: when At that time, a constant voltage charging command is output, and the charging voltage is... ; Trickle mode: when At that time, a trickle charging command is output, and the charging current is... ; Path planning decision: Based on the improved A* algorithm, using path cost With the goal of minimizing the minimum, plan the optimal driving path and output the path execution command. The path planning formula is: ,in The cost from the starting node to the current node. This is the heuristic function from the current node to the target node; Operation and maintenance early warning judgment: predicting the probability of failure based on LSTM neural network model ,when When, output operation and maintenance early warning commands; when When this happens, an emergency stop command is output.

[0013] Preferably, the execution control steps specifically include: Motion control: Based on path planning instructions, control the chassis drive module to perform driving operations, including front wheel steering angle. driving speed Dynamically adjust based on path complexity, straight line segment Turning section Obstacle crossing section ; Dynamic adjustment of charging parameters: Based on charging strategy instructions, the output voltage and current are adjusted through the power module. Voltage adjustment formula. ,in For the vehicle's required voltage, For optimal temperature, , This is the adjustment coefficient; Battery protection control: Based on safety status commands, the corresponding protection mechanism is activated. Overvoltage occurs, the charging circuit is cut off; overtemperature occurs, the liquid cooling system is activated, and the coolant flow rate is adjusted. Disconnect the high-voltage box in case of an insulation fault; Cloud-based interactive control: The charging data and device status data are uploaded to the cloud platform via the 4G / WIFI communication module, and the scheduling instructions and algorithm update instructions from the cloud platform are received. Data transmission adopts an encrypted protocol. Human-computer interaction control: The touch screen displays information such as charging progress, SOC, and charging cost. It receives user's QR code scanning, account password, and card payment instructions. Charging starts after payment verification and a charging report is generated and displayed after payment is completed.

[0014] Preferably, the feedback optimization step specifically includes: Collect feedback data: including charging efficiency SOC estimation error SOH estimation error Charging time deviation Number of equipment failures; Algorithm optimization: Adjusting the parameters of the feature extraction algorithm based on gradient descent, and optimizing the process noise covariance of EKF. Observation noise covariance The formula is: in, For parameters to be optimized, For learning rate, Let it be the error loss function; Model update: Input feedback data into the decision model for retraining, update model parameters, and adjust fit weights. The weight matrix of the fault prediction model; Parameter storage: The optimized algorithm parameters and model parameters are stored in the flash memory of the monitoring core board and automatically loaded on the next startup.

[0015] An intelligent control system for a mobile energy storage and charging robot, the system comprising a hardware layer and a software layer, the hardware layer and the software layer being connected via a bus communication connection: Hardware layer: Monitoring core board: adopts automotive-grade STM32H743VI chip, integrating CPU, memory, flash memory, CAN bus interface, RS232 serial port, and Ethernet interface to realize system monitoring, coordination control, data processing and cloud interaction functions; Energy storage system: includes lithium iron phosphate battery pack, high voltage box, and BMS. The BMS uses TI's BQ79616 chip to realize battery status monitoring, equalization control and safety protection. Power system: includes DC / DC power module, filter unit, and metering unit, wherein the power module adopts forced air cooling; Chassis drive module: includes Ackermann chassis, drive motor, steer-by-wire EPS system, EHB electro-hydraulic braking system, and front and rear coil spring suspension, to realize robot movement control; Communication module: integrates 4G, WIFI and Bluetooth modules to enable communication with the cloud platform, the vehicle to be charged and the human-machine interaction module; Sensor group: includes voltage sensor, current sensor, temperature sensor, GPS positioning module, lidar, and ultrasonic sensor; Human-computer interaction module: including touch screen, QR code scanning module, card swiping module, status indicator light, and advertising screen; Protection module: includes overvoltage protection circuit, overcurrent protection circuit, overtemperature protection circuit, insulation detection circuit, and emergency stop button; Software layer: Data acquisition module: drives the sensor group, BMS, and power module to acquire data from multiple sources, and realizes data caching and synchronization; Data preprocessing module: integrates Kalman filtering, min-max normalization, 3σ anomaly detection, and linear interpolation functions to process the raw data; Feature extraction module: Extracts battery status features, charging demand adaptation features, environmental adaptation features, and location scheduling features; Decision control module: includes compatibility judgment unit, security judgment unit, policy matching unit, path planning unit, and operation and maintenance early warning unit, and outputs decision instructions; Execution control module: Drives the chassis drive module, power module, BMS, and communication module to perform control operations; Cloud interaction module: Enables data uploading, command reception, algorithm updates, and remote operation and maintenance with the cloud platform; Operation and maintenance optimization module: Based on feedback data, optimize algorithms and model parameters to generate operation and maintenance reports.

[0016] Preferably, the system further includes a cloud platform interaction unit, the cloud platform comprising: Data storage module: Stores robot status data, charging records, user data, and operation and maintenance data, using a distributed database; Scheduling Management Module: Implements multi-robot collaborative scheduling based on a greedy algorithm. The scheduling objective function is: ; in The charging revenue for the i-th robot, For energy consumption costs, As scheduling priority weights; Operation and maintenance management module: Generates inspection plans based on big data analysis, predicts equipment failures, and outputs operation and maintenance instructions; User management module: Manages user accounts, payment information, and charging records, and supports WeChat, Alipay, and UnionPay payments; Algorithm update module: Sends algorithm update packages to the robot, supporting OTA remote upgrades.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention adapts to all national and European standard vehicle models through charging demand adaptation calculation and dynamic protocol matching, and is compatible with different types of batteries such as lithium iron phosphate and ternary lithium, thus solving the problem of poor compatibility in existing technologies. It adopts the SOC estimation method of ampere-hour integral + EKF and the SOH estimation model of nonlinear regression, combined with battery consistency equalization control, to improve battery management accuracy and extend battery life. This invention dynamically adjusts the charging strategy based on battery status and environmental parameters, and uses a DC / DC power module to achieve high-efficiency energy conversion. The full-load charging efficiency is ≥95%, and 60-120 kWh can be charged in 1 hour. The charging power fluctuation is ≤5% in extreme environments, which solves the problems of low charging efficiency and insufficient environmental adaptability of existing technologies. Based on the improved A* algorithm, optimal path planning is achieved. Combined with multi-robot collaborative scheduling on a cloud platform, the energy consumption of path planning is reduced by 20%, and the average waiting time of vehicles is shortened by more than 30%, which solves the problem of unintelligent scheduling in existing technologies. This invention also achieves fault prediction based on LSTM neural networks, with a fault prediction accuracy of ≥90%, proactive maintenance response time of ≤30min, and a 50% improvement in mean time between failures (MTBF), thus reducing maintenance costs. It supports multiple operation methods such as scanning codes, swiping cards, and account passwords, as well as multi-channel payments via WeChat / Alipay / UnionPay. Charging information is visualized in real time, and the user operation success rate is ≥99.5%, improving the user experience. The system adopts a modular design, with the hardware layer supporting the expansion of components such as sensors and power modules, and the software layer supporting OTA remote upgrades of algorithms and models, adapting to the future technological development needs of new energy vehicles. Attached Figure Description

[0018] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein: Figure 1 This is an overall architecture diagram of an intelligent control system for a mobile energy storage and charging robot according to the present invention. Figure 2 This is an overall flowchart of an intelligent control method for a mobile energy storage and charging robot according to the present invention. Detailed Implementation

[0019] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0020] Specific embodiments of the present invention are described below with reference to the accompanying drawings.

[0021] Please see Figures 1-2 This embodiment proposes an intelligent control system and method for a mobile energy storage and charging robot; The system of this invention takes the monitoring core board as the control center, integrates the hardware resources of the energy storage system, power system, chassis drive module, BMS, communication module, sensor group and human-machine interaction module, and combines the software functions of data acquisition, preprocessing, feature extraction, decision control, cloud interaction and operation and maintenance optimization to realize the intelligent control of the mobile energy storage charging robot throughout the entire process. The method of this invention uses a robot as the execution subject and follows the computer data processing logic of data acquisition, data processing, feature extraction, decision judgment, execution control and feedback optimization. Through multi-source data fusion, intelligent algorithm modeling and dynamic decision control, it realizes charging compatibility and adaptation, precise battery management, intelligent scheduling, efficient charging and cloud collaborative operation and maintenance.

[0022] The intelligent control system of this invention comprises a hardware layer and a software layer. The hardware layer provides physical support for the system, while the software layer implements data processing and intelligent decision-making. The two layers interact with each other through communication interfaces such as CAN bus, RS232 serial port, and Ethernet. The specific components are as follows: The hardware layer is the physical foundation of the intelligent control system, including the monitoring core board, energy storage system, power system, chassis drive module, communication module, sensor group, human-machine interaction module, protection module, etc., as well as the structure, parameters and connection relationships of each component.

[0023] The monitoring core board is the heart of the entire system, responsible for data processing, logical judgment, device control, and cloud interaction. Its specific parameters are as follows: Main chip: Automotive-grade STM32H743VI, ARM Cortex-M7 core, 480MHz; Memory: 512KB RAM + 2MB Flash, expandable to 16MB Flash; Communication interfaces: 2 CANFD interfaces (supporting CAN2.0A / B protocol), 4 RS232 serial ports, 1 Ethernet interface (RJ45), and 1 USB3.0 interface; Power supply: DC12V input, power consumption ≤5W; Operating temperature: -40℃~85℃, protection rating IP54; Connection relationships: Connects to BMS, power module, and chassis drive module via CAN bus; connects to sensor group and human-machine interaction module via RS232 serial port; connects to communication module via Ethernet interface to achieve communication with cloud platform.

[0024] The core functions of the monitoring core board include: Data processing: Receives and processes data collected from devices such as sensor arrays, BMS, and power modules, and performs preprocessing and feature extraction algorithms; Decision control: Run the decision model to generate instructions such as charging strategy, path planning, and protection control; Equipment control: Sends control commands to power modules, chassis drive modules, BMS, etc., to control equipment operation; Cloud Interaction: Data is uploaded and commands are received with the cloud platform through the communication module, supporting OTA remote upgrades; Fault handling: Detect equipment faults, trigger protection mechanisms, generate fault reports, and upload them to the cloud platform.

[0025] The energy storage system is the robot's energy source, including lithium iron phosphate battery packs, high-voltage boxes, and a BMS. Specific parameters are as follows: ; The core functions of an energy storage system include: Energy storage: Store electrical energy by supplementing power through external charging piles or the power grid; Condition monitoring: The BMS monitors parameters such as cell voltage, current, temperature, SOC, and SOH in real time; Balance control: When the voltage difference between individual cells is ≥0.05V, passive or active balance is activated to ensure battery consistency; Safety protection: Provides protection against overvoltage, overcurrent, overtemperature, undervoltage, and insulation faults, and disconnects the high-voltage circuit when the protection is triggered.

[0026] The power system is responsible for converting the electrical energy from the energy storage battery into the voltage and current required by the vehicle being charged. It includes a DC / DC power module, a filtering unit, and a metering unit. Specific parameters are as follows: ; The core functions of a power system include: Voltage-to-current conversion: Converting the high-voltage DC power from the energy storage battery into the voltage and current required by the vehicle being charged; Power regulation: Dynamically adjusts output power, voltage, and current based on decision commands; Metering and Statistics: Accurately measure charging power to provide a basis for billing; Electromagnetic compatibility: Electromagnetic interference is suppressed through the filtering unit to ensure stable operation of the equipment.

[0027] The chassis drive module enables the robot's movement and includes an Ackerman chassis, drive motors, steering system, braking system, and suspension system. Specific parameters are as follows: ; The core functions of the chassis drive module include: Motion control: Based on path planning instructions, control the motor speed and steering angle to achieve actions such as straight-line driving, turning, and obstacle crossing; Speed ​​adjustment: Dynamically adjusts driving speed according to road conditions to ensure smooth and safe driving; Braking control: Receives braking commands to achieve emergency braking or deceleration braking; Attitude monitoring: The chassis attitude is monitored by gyroscopes and accelerometers to prevent rollover.

[0028] The communication module enables communication between the robot and the cloud platform, the vehicle to be charged, and the human-machine interaction module. It includes a 4G module (supporting TD-LTE / FDD-LTE), a Wi-Fi module (supporting 802.11n), and a Bluetooth module. Specific parameters are as follows: ; The core functions of the communication module include: Cloud platform communication: Upload robot status data, charging records, and fault information; receive scheduling instructions and algorithm update instructions. Vehicle communication: Communicates with the BMS of the vehicle to be charged to obtain vehicle charging demand data and provide feedback on charging status; Local communication: Communicates with the human-machine interaction module to transmit operation commands and display data.

[0029] The sensor array is used to collect multi-source data, including voltage sensors, current sensors, temperature sensors, GPS positioning modules, LiDAR, and ultrasonic sensors. Specific parameters are as follows: ; The core functions of the sensor array include: Status data acquisition: Collects battery voltage, current, temperature, power module status, chassis motor status, etc. Environmental data acquisition: Collecting environmental data such as temperature, humidity, altitude, atmospheric pressure, and obstacle information; Location data acquisition: Collect the robot's own position coordinates, the position coordinates of the vehicle to be charged, etc. Attitude data acquisition: Collects chassis attitude angles, steering angles, etc., for driving control.

[0030] The human-computer interaction module enables users to interact with the robot, including a touch screen, a QR code scanning module, a card swiping module, status indicator lights, and an advertising screen. Specific parameters are as follows: ; The core functions of the human-computer interaction module include: Operation input: Receive user operation instructions such as scanning QR codes, swiping cards, and logging in with account passwords; Information display: Displays charging progress, SOC, charging cost, device status, fault prompts, and other information; Payment interaction: Supports WeChat Pay, Alipay, and UnionPay payments, displays payment QR code, and provides payment result feedback; Advertising Placement: Play advertising content and support remote updates of advertising materials.

[0031] The protection module ensures the safe operation of the robot and includes overvoltage protection circuit, overcurrent protection circuit, overtemperature protection circuit, insulation detection circuit, and emergency stop button. Specific parameters are as follows: ; The core functions of the protection module include: Real-time monitoring: Monitor parameters such as voltage, current, temperature, and insulation resistance to determine whether they exceed safety thresholds; Protection Trigger: When a parameter exceeds the threshold, the corresponding protection mechanism is immediately triggered to prevent equipment damage or safety accidents; Fault alarm: When the protection is triggered, an alarm is issued through the status indicator light and the touch screen, and the data is uploaded to the cloud platform at the same time.

[0032] The software layer runs on the monitoring core board and cloud platform, including a data acquisition module, a data preprocessing module, a feature extraction module, a decision control module, an execution control module, a cloud interaction module, and an operation and maintenance optimization module. The functions and implementation methods of each module are as follows: The data acquisition module is responsible for driving the hardware device to acquire data from multiple sources, realizing data caching, synchronization, and preliminary processing. Its functional flow is as follows: Device driver: Load the drivers for devices such as sensor groups, BMS, power modules, and chassis drive modules, and establish communication connections; Data Acquisition Scheduling: Based on the preset acquisition frequencies (battery parameters 10Hz, environmental parameters 1Hz, location parameters 5Hz, device status parameters 2Hz), each device is scheduled to acquire data. Data caching: A circular buffer is used to store the collected data. The buffer size is 1024KB and supports data overwriting (oldestdatafirst). Data synchronization: Data collected from different devices is synchronized based on timestamps, with a synchronization error of ≤1ms; Data verification: Perform CRC verification on the collected data. If the verification fails, discard the data and re-collect the data.

[0033] The data preprocessing module performs noise reduction, standardization, anomaly detection, and data completion on the collected raw data to improve data quality and provide reliable data for subsequent feature extraction and decision-making. The implementation methods of each processing step are as follows: Noise Reduction: The Kalman filter algorithm is used to reduce noise in high-frequency fluctuation data such as battery voltage, current, and temperature. The filter parameters are dynamically adjusted according to the data type. For example, the process noise covariance of battery voltage is Q=diag([1e-6,1e-6]), and the observation noise covariance is R=1e-4. Standardization processing: min-max standardization is used to normalize data such as voltage range and power demand, mapping the data to the [0,1] interval to eliminate the influence of dimensions; Anomaly detection: The 3σ criterion is used to remove outliers from the data. The mean μ and standard deviation σ of the data are calculated. When the data satisfies |x-μ|>3σ, it is determined to be an outlier and replaced with valid data from the previous time step. Data completion: Linear interpolation is used to complete missing data. When the missing data time is ≤1s, linear interpolation is used; when the missing time is >1s, a data missing alarm is issued.

[0034] The feature extraction module extracts multi-dimensional feature data based on preprocessed standardized data, including battery state features, charging demand adaptation features, environmental adaptation features, and location scheduling features. The extraction methods for each feature are as follows: Battery state feature extraction: SOC characteristics: The SOC is estimated by combining the ampere-hour integration method with EKF. The ampere-hour integration method calculates the preliminary SOC, and EKF corrects the integration error based on the battery voltage observation value to improve the estimation accuracy. SOH characteristics: SOH is estimated based on a nonlinear regression model. The model parameters are obtained by fitting a large amount of battery cycle experimental data. The least squares method is used for fitting. Battery consistency characteristics: Calculating the standard deviation of individual cell voltages and temperature standard deviation Consistency coefficient ,in This refers to the battery's nominal voltage. Charging demand adaptation feature extraction: Voltage range matching When the vehicle voltage range With robot output voltage range When the intersection is not empty, ;otherwise ; Power demand matching : ,in This represents the robot's maximum output power. Charging protocol compatibility By parsing the vehicle's charging protocol version and matching it with the protocol library supported by the robot, if a match is successful... ,otherwise ; Charging protocol compatibility By parsing the vehicle's charging protocol version and matching it with the protocol library supported by the robot, if a match is successful... ,otherwise ; Fit F: Calculated using a weighted summation formula, with weighting coefficients... , , (Obtained through machine learning optimization); Environmental adaptation feature extraction: Temperature adaptability coefficient When the ambient temperature hour, ;when or hour, ; Humidity adaptability coefficient ; Altitude Adaptability Coefficient ; Environmental adaptability coefficient ; Location scheduling feature extraction: Path length feature: Calculate the robot's current position coordinates based on GPS positioning data. The straight-line distance between them: ; Path cost characteristics: Taking into account driving distance, gradient, obstacle crossing conditions, and energy consumption, path cost... ,in: d represents the actual planned path length (m), calculated by fusing LiDAR and GPS data. s is the slope coefficient If the percentage exceeds 20%, it is considered an impassable path; o represents the obstacle clearance coefficient, and a single 70mm speed bump. For every 10mm increase in obstacle clearance height, 0.2 is added; if the obstacle clearance height exceeds 100mm, it is deemed impassable. e represents the estimated energy consumption (kWh). Driving time (s); Weight system The genetic algorithm was used to optimize and determine the outcome.

[0035] The decision control module is the core of the software layer. Based on the multi-dimensional feature data output by the feature extraction module, it uses a preset decision model and algorithm to perform functions such as charging compatibility judgment, battery safety status judgment, charging strategy matching judgment, path planning judgment, and operation and maintenance early warning judgment, and outputs corresponding decision instructions. The implementation methods of each judgment unit are as follows: Charging compatibility assessment: Input: Charging demand fit F; Decision-making logic: When If the system is deemed fully compatible, a "Charging Allowed" command will be output, and the charging power will be executed according to the maximum power required by the vehicle or the maximum value of the robot's rated power. If the system is determined to be partially compatible, a "reduced power charging" command will be output, where the charging power = maximum power required by the vehicle × F. If the system determines that the vehicle is incompatible, it outputs a "reject charging" command and prompts the user through the human-machine interaction module that "the current vehicle does not support charging". Output: Compatibility assessment results and corresponding charging allow / restriction instructions.

[0036] Battery safety status assessment: Input: Battery status ; Decision-making logic: Overvoltage judgment: Set individual cell Based on battery cell parameters), when any single cell... It outputs an "overvoltage protection" command to cut off the charging circuit; Output a "buck charging" command, reducing the charging voltage by 5%; Over-temperature detection: Set to Level 1 Issue a "preheat start" command to activate the battery preheating module; Insulation assessment: Set insulation resistance threshold It outputs an "insulation fault" command, cuts off the high-voltage circuit, and simultaneously issues an audible and visual alarm. Consistency judgment: When the consistency system The "Start Balancing" command activates the BMS's active balancing function, balancing the power supply. It outputs a "limit charging" command, reducing the charging current by 30%. Output: Battery safety status assessment result and corresponding protection control commands.

[0037] Charging strategy matching judgment: Inputs: Battery SOC, Battery Temperature Vehicle charging requirements (target SOC, charging time requirements); Decision-making logic: Constant current mode: Voltage; Constant pressure mode: ; Trickle mode (C represents the vehicle battery's rated capacity). Fast charging mode: When the vehicle charging time is short... At that time, output a "fast charging" command to charge. Enhance liquid cooling; Output: Charging mode command and corresponding charging parameters (voltage, current).

[0038] Path planning judgment: Input: Robot's current position, target parking space position, environmental obstacle information, chassis parameters (maximum gradeability, obstacle crossing height); Decision logic: Optimal path planning is achieved based on the improved A* algorithm. The specific steps are as follows: Step 1: Create a grid map, divide the driving area into 10cm×10cm grids, and mark the obstacle grids (the grids corresponding to obstacles detected by the LiDAR). Step 2: Define the starting node (the grid corresponding to the robot's current position) and the target node (the grid corresponding to the target parking space position); Step 3: Calculate node generation The actual path cost (calculated based on the path cost formula) The heuristic function for a point uses Euclidean distance. ; Step 4: Expand nodes using a priority queue (min-heap), selecting... The smaller node becomes the next expansion node; Step 5: Repeat steps 3-4 until the target node is found and the optimal path is generated; Step 6: Path smoothing. B-spline curves are used to smooth the planned path, avoiding sharp turns. Output: Optimal path coordinate sequence and corresponding driving distance. barrier ).

[0039] Operation and maintenance early warning judgment: Input: Equipment status characteristics (power module temperature, chassis motor current, communication signal strength, battery degradation rate), historical fault data; Decision-making logic: Based on LSTM neural network model prediction The specific implementation is as follows: LSTM network structure: Input layer (6 neurons, corresponding to 6 device state features) → Hidden layer (2 layers, 32 neurons per layer) → Output layer (1 neuron, outputting the fault probability). Training data: Collect 1000 sets of normal equipment operation data and 500 sets of fault data, and divide them into training set and test set in a 7:3 ratio; Activation functions: The hidden layer uses the ReLU function, and the output layer uses the Sigmoid function; Loss function: Cross-entropy loss function Predicting probabilities; Optimizer: Adam optimizer, learning rate = 0.001, number of iterations = 1000; Decision threshold It outputs an "emergency stop" command to immediately stop charging and movement operations; Output: Operation and maintenance early warning results and corresponding shutdown / early warning instructions.

[0040] The execution control module receives decision commands from the decision control module and drives the chassis drive module, power module, BMS, communication module, human-machine interface module, and other devices in the hardware layer to perform corresponding control operations, ensuring accurate execution of the commands. The implementation methods of each control unit are as follows: Movement control: Input: Path planning instructions (optimal path coordinate sequence, driving speed instructions); Control logic: Steering control: Calculate the steering angle based on the coordinate difference between the current position and the next path node. ,in The front steering angle is controlled by the steer-by-wire EPS steering system, with a steering response time of ≤0.2s. Speed ​​control: Adjust the driving speed according to the type of path segment (straight section, turning section, obstacle crossing section), and control the speed of the chassis drive motor through PWM signal, with a speed control accuracy of ±0.2km / h; Braking control: When encountering a sudden obstacle (the lidar detects an obstacle within 1m) or receiving a braking command, the EHB electronic hydraulic braking system is activated, with a braking deceleration of ≥2m / s² and a braking distance of ≤0.4m (at a vehicle speed of 5km / h). Outputs: Steering angle control signal, motor speed control signal, braking control signal.

[0041] Dynamic adjustment of charging parameters: Input: Charging strategy instructions (charging mode, voltage, and current parameters); Control logic: Voltage regulation: The output voltage is regulated by the PWM controller of the DC / DC power module, with a voltage regulation accuracy of ≤ ±0.5% and a regulation response time of ≤ 1s; Current regulation: The output current is regulated by the current limiting circuit of the power module. The current regulation accuracy is ≤±1%, and the current limiting value adjustment range is 0.5A~250A. Power regulation: The output power is calculated in real time based on the product of voltage and current. When the power exceeds 105% of the rated power, the current is automatically reduced to ensure power stability. Outputs: DC / DC module voltage control signal and current control signal.

[0042] Battery protection control: Input: Battery safety status commands (overvoltage, overtemperature, insulation fault, and other protection commands); Control logic: Overvoltage protection: Controls the contactor in the high-voltage box to disconnect, cutting off the connection between the battery and the power module. The disconnection response time is ≤10ms. Over-temperature protection: Controls the start of the water pump and cooling fan of the liquid cooling system, regulates the coolant flow, and cuts off the charging circuit when the temperature continues to rise; Insulation protection: The insulation monitoring relay inside the control high-voltage box is disconnected, the auxiliary power supply is cut off, and an audible and visual alarm is triggered. Equalization control: The active equalization circuit of the control BMS is activated to transfer energy from cells with higher voltage to cells with lower voltage. The equalization time is ≤1h (cell voltage difference ≤0.05V). Outputs: contactor control signal, liquid cooling system control signal, equalization circuit control signal, alarm control signal.

[0043] Cloud-based interactive control: Input: Cloud platform interaction commands (scheduling commands, algorithm update commands, operation and maintenance commands); Control logic: Data Upload: Robot status data, charging records, and fault information are uploaded to the cloud platform at a preset frequency (status data once every 10 seconds, charging record once per charging completion, and fault information uploaded in real time). Data transmission uses the MQTT protocol, and data encryption uses the AES-256 algorithm. Command reception: Real-time monitoring of commands issued by the cloud platform, with a command reception delay of ≤1s. After reception, CRC verification is performed, and the command is parsed and executed after the verification is passed. OTA upgrade: When an algorithm update command is received, the update package is downloaded via the 4G / WIFI module. After the update package is verified, the software program of the monitoring core board is updated. The charging function is kept normal during the upgrade process, and the upgrade success rate is ≥99%. Output: Uploaded data, command response signal, upgrade status signal.

[0044] Human-computer interaction control: Input: Human-computer interaction commands (user operation commands, display commands); Control logic: Operation response: Receive user's login instructions via QR code scanning, card swiping, or account password, perform identity verification and payment verification, and start charging after successful verification. The verification response time is ≤1 second. Information display: The touch screen displays charging progress (real-time SOC value, charging time, remaining time), charging cost (real-time cost, total cost), and device status (normal, warning, fault), with an update frequency of 1 time / 1 second; Payment interaction: Generates WeChat, Alipay, and UnionPay payment QR codes, receives payment result notifications from payment platforms, and automatically starts charging after payment is completed; payment success rate ≥99.5%. Outputs: Touchscreen display signal, barcode scanning module driver signal, card swiping module driver signal, payment QR code generation signal.

[0045] The cloud interaction module enables two-way data interaction between the robot and the cloud platform, including functions such as data uploading, command receiving, algorithm updates, and remote operation and maintenance. Its core functions are as follows: Data upload function: Uploaded data types include: robot status data (battery status, power module status, chassis status, environmental parameters), charging record data (charging time, charging amount, charging cost, user information), fault data (fault type, fault time, fault parameters), and operation and maintenance data (inspection records, maintenance records). Upload frequency: Status data once every 10 seconds, charging record data once per charging completion, fault data uploaded in real time, and maintenance data uploaded once every 24 hours. Data format: Encapsulated in JSON format, data fields include device ID, timestamp, data type, data value, and verification code; Transmission protocol: MQTT protocol is used, supporting QoS Level 1 message transmission (ensuring that messages are delivered at least once).

[0046] Command receiving function: Types of instructions received: scheduling instructions (target parking space allocation, charging priority adjustment), algorithm update instructions (feature extraction algorithm parameter update, decision model weight update), operation and maintenance instructions (inspection tasks, fault handling instructions), parameter configuration instructions (charging voltage range adjustment, safety threshold modification); Command parsing: The received command is parsed using JSON and checked for CRC. If the check passes, the command content is extracted; if the check fails, a resend request is made. Command Response: After executing the command, a response message is sent to the cloud platform, including the command execution result (success / failure), execution time, and current status.

[0047] OTA remote upgrade function: Upgrade process: Cloud platform issues upgrade notification → Robot downloads upgrade package → Upgrade package verification → Algorithm / program update → Upgrade result feedback; Upgrade package features: Differential upgrade packages are used to reduce the amount of data transmitted, and the upgrade packages are encrypted during transmission (RSA+AES) to ensure security; Upgrade Guarantee: If a network interruption occurs during the upgrade process, the download can continue after the network is restored; if the upgrade fails, it will automatically roll back to the original version to ensure the robot operates normally.

[0048] Remote operation and maintenance functions: Remote monitoring: Receives real-time status data uploaded by the robot and displays the device's operating status, charging progress, and fault information on the cloud platform interface; Remote diagnostics: Automatically analyzes fault data to identify the cause of the fault, generates a diagnostic report, and provides fault handling suggestions; Remote control: Supports remote start / stop charging, remote adjustment of charging parameters, remote restart of the robot, etc., with a remote control response time of ≤3s.

[0049] The operation and maintenance optimization module, based on the robot's operational and feedback data, iteratively optimizes the feature extraction algorithm, decision model, and control parameters to improve the system's stability, accuracy, and efficiency. Its core functions are as follows: Feedback data collection: Data types collected: OC estimation error Charging time deviation Number of failures and user operation success rate; Data collection frequency: Once a data is collected after each charge is completed; real-time data collection is performed in case of device failure. Data storage: Feedback data is stored in the Flash memory of the monitoring core board, with a storage capacity of ≥1000 records. Once the capacity is exceeded, the oldest data is overwritten by timestamp.

[0050] Algorithm parameter optimization: Feature extraction algorithm optimization: Adjusting the process noise covariance Q and observation noise covariance R of EKF based on gradient descent method, target study Number of times = 100; Decision Model Weight Optimization: Optimizing the Weights of Charging Demand Fit Based on Genetic Algorithm Letter of Inquiry Population size = 50, number of iterations = 50, crossover probability = 0.8, mutation probability = 0.1; Control parameter optimization: Optimizing PID parameters for speed control, voltage control, and current control based on PID parameter self-tuning algorithm. The tuning method adopts the critical proportional method, and the error convergence time is controlled to be ≤2s.

[0051] Fault prediction and operation and maintenance planning: Fault prediction: Based on the LSTM neural network model, input device status feature data, predict the failure probability in the next 7 days, and generate operation and maintenance early warning information when the failure probability is ≥0.7; Operation and maintenance planning: Based on equipment uptime, fault prediction results, and peak charging period distribution, an optimal inspection plan is generated for inspection. The algorithm uses a greedy approach to minimize the total inspection distance.

[0052] Operation and maintenance report generation: Report content includes: Equipment operation status statistics (cumulative charging times, cumulative charging capacity, average charging efficiency), fault statistics (fault type, number of faults, fault cause), algorithm optimization effect (comparison of errors before and after optimization), and operation and maintenance suggestions (inspection time, maintenance items, parameter adjustment suggestions). Generation frequency: Once a week, the generated data is uploaded to the cloud platform. Report format: PDF, including data tables, trend charts, and optimization curves.

[0053] Please continue reading. Figures 1-2 The intelligent control method of this invention uses a mobile energy storage and charging robot as the execution subject, and follows the computer data processing logic of "data acquisition - data processing - feature extraction - decision judgment - execution control - feedback optimization", the specific process of which is as follows: Figure 1 As shown, the detailed steps are as follows: Step 1: System Initialization After the robot is powered on, the monitoring core board starts up, initializes various hardware modules (energy storage system, power system, chassis drive module, communication module, sensor group, human-machine interaction module), loads the software layer's algorithms, models and parameters (reading preset parameters from Flash), establishes an MQTT connection with the cloud platform, completes system self-test (hardware status, communication status, battery status detection), and enters standby mode after passing the self-test, waiting for user charging requests or cloud platform scheduling instructions.

[0054] Step 2: Data Collection When the robot receives a user's charging request (triggered by scanning a code, swiping a card, or logging in with an account and password) or a cloud platform scheduling instruction (assigning a target parking space), it initiates the data collection process: The robot collects external environmental data through a sensor array: ambient temperature, humidity, altitude, atmospheric pressure, and obstacle information (location, size, and type). The system collects its current location coordinates using a GPS positioning module, and detects surrounding obstacles using lidar and ultrasonic sensors to generate an environmental map. The system communicates with the BMS of the vehicle to be charged via a communication module to collect vehicle charging demand data, including: battery type (lithium iron phosphate / ternary lithium, etc.), rated capacity, voltage range, charging power demand, initial SOC, target SOC, charging protocol version (Chinese standard / European standard), and vehicle location coordinates. The BMS collects state data of its own energy storage batteries (cell voltage, total voltage, charge / discharge current, cell temperature, total temperature, initial SOC (State of Charge), initial SOH (State of Health), insulation resistance, coolant flow rate and temperature). Data such as output voltage, output current, module temperature, and operating efficiency are collected through the power module. The chassis drive module collects data such as chassis motor status (speed, current, voltage), steering angle, and driving speed. The communication module receives scheduling instructions (scheduling priority, peak and off-peak time period identifiers, and charging price information) issued by the cloud platform.

[0055] Step 3: Data Preprocessing The robot preprocesses the collected initial data from multiple sources to eliminate the effects of noise, outliers, and missing data. Kalman filtering algorithm is used to reduce noise in high-frequency fluctuation data such as battery cell voltage, current, and temperature, thereby reducing measurement noise interference. Min-max normalization is used to normalize data of different dimensions such as voltage range and power demand, and map them to the [0,1] interval; The 3σ criterion was used to detect outliers in all data, removing outliers that exceeded three times the standard deviation and replacing them with valid data from the previous time step. Linear interpolation is used to complete missing data to ensure data integrity. Perform CRC check on the preprocessed standardized data to ensure the accuracy of data transmission.

[0056] Step 4: Feature Extraction The robot extracts multi-dimensional feature data based on standardized data using a pre-set feature extraction algorithm: Extracting battery state characteristics: including SOC (estimated based on ampere-hour integral + EKF), SOH (estimated based on nonlinear regression model), and battery consistency coefficient (calculated based on the standard deviation of single cell voltage). Extract charging demand adaptation features: including voltage range matching degree, power demand matching degree, and charging protocol adaptation degree, and calculate the comprehensive adaptation degree F; Extract environmental adaptation features: including temperature adaptation coefficient, humidity adaptation coefficient, and altitude adaptation coefficient; calculate the environmental adaptation coefficient. ; Extract location scheduling features, including path length features (straight-line distance) and path cost features (combined distance, gradient, obstacle crossing, and energy consumption).

[0057] Step 5: The decision-making robot inputs multi-dimensional feature data into the decision-making model to perform multi-dimensional decision-making: Charging compatibility assessment: Determine whether charging the current vehicle is supported based on the compatibility degree F, and output "Allow charging", "Charge at reduced power" or "Reject charging" commands. Battery safety status assessment: Based on the battery status characteristics, determine whether there are safety hazards such as overvoltage, overtemperature, insulation failure, and poor consistency in the battery, and output corresponding protection and control commands; Charging strategy matching judgment: Based on battery SOC, temperature and vehicle charging needs, select constant current, constant voltage, trickle or fast charging mode, and determine charging voltage and current parameters. Path planning and judgment: Based on the improved A* algorithm, with the goal of minimizing path cost, the optimal driving path from the current position to the target parking space is planned, and the driving speed of each path segment is determined; Operation and maintenance early warning judgment: Based on the LSTM neural network model, predict the probability of equipment failure and output "operation and maintenance early warning", "emergency shutdown" or "normal operation" instructions.

[0058] Step 6: Execute Control Based on the instructions output by the decision-making process, the robot drives each hardware module to perform corresponding control operations: Motion control: Based on path planning instructions, control the steering, speed and braking of the chassis drive module to achieve precise movement of the robot towards the target parking space, and avoid obstacles in real time during the movement; Charging preparation: After the robot arrives at the target parking space, it prompts the user to connect the charging gun through the human-computer interaction module and checks the connection status of the charging gun (a confirmation signal is sent after successful connection). Charging parameter adjustment: According to the charging strategy command, control the DC / DC power module to adjust the output voltage and current to start charging; Battery protection control: The battery status is monitored in real time during charging, and the corresponding protection mechanism is activated when safety hazards such as overvoltage or overtemperature occur. Cloud interactive control: Uploads charging progress and device status data to the cloud platform in real time and receives real-time instructions from the cloud platform; Human-computer interaction control: The system displays charging progress, SOC, charging cost and other information in real time via a touch screen, receives user payment instructions, and generates a charging report after payment is completed.

[0059] Step 7: Feedback Optimization After charging is complete or equipment malfunction is resolved, the robot initiates a feedback optimization process: Collect feedback data, including charging efficiency, SOC estimation error, SOH estimation error, charging time deviation, fault information, and user operation success rate. Algorithm parameter optimization: Based on feedback data, gradient descent, genetic algorithm and other methods are used to optimize the feature extraction algorithm parameters and decision model weights; Control parameter optimization: PID parameters for speed, voltage, and current control are optimized through PID parameter self-tuning; Fault prediction and operation and maintenance planning: Based on the optimized model, predict the probability of equipment failure and generate operation and maintenance early warning and inspection plans; Parameter storage and update: The optimized parameters are stored in Flash and automatically loaded on the next startup, while also being uploaded to the cloud platform for backup.

[0060] Step 8: System Standby / Power Off After feedback optimization is completed, the robot disconnects the charging gun and returns to the charging base station to recharge or proceeds to the next target parking space according to cloud platform instructions or user operations. If there is no task for a long time (more than 2 hours), it enters a low-power standby state with a standby power consumption of ≤10W. If a shutdown command is received, the power supply of each hardware module is turned off to complete the shutdown.

[0061] The intelligent control system and method for mobile energy storage and charging robots of the present invention have the following significant technical advantages compared with the prior art: Strong compatibility: Through charging demand adaptation calculation and dynamic protocol matching, it is compatible with all national standard (GB / T18487.1-2015) and European standard (CE) models, and is compatible with different types of batteries such as lithium iron phosphate and ternary lithium. The voltage output range is DC150-1000V, and the adaptation rate is ≥98%, which solves the problem of poor compatibility of existing technologies. Precise battery management: Employing an ampere-hour integral + EKF SOC estimation method and a nonlinear regression SOH estimation model, combined with battery consistency equalization control, the SOC estimation error is ≤ ±1.5%, the SOH estimation error is ≤ ±2%, and the battery consistency coefficient is ≥ 0.98, significantly improving battery management accuracy and extending battery life; High charging efficiency: The charging strategy is dynamically adjusted based on battery status and environmental parameters. A DC / DC power module is used to achieve efficient energy conversion. The full-load charging efficiency is ≥95%, and 60-120KWH can be charged in 1 hour. The charging power fluctuation is ≤5% under extreme environments (-20℃~50℃), which solves the problems of low charging efficiency and insufficient environmental adaptability of existing technologies. Intelligent scheduling: Based on the improved A* algorithm to achieve optimal path planning, combined with multi-robot collaborative scheduling on the cloud platform, the energy consumption of path planning is reduced by 20%, and the average waiting time of vehicles is shortened by more than 30%, which solves the problem of unintelligent scheduling in existing technologies. Highly efficient operation and maintenance: Fault prediction is achieved based on LSTM neural network, with a fault prediction accuracy of ≥90%, proactive operation and maintenance response time of ≤30min, and the mean time between failures (MTBF) of equipment is increased by 50%, reducing operation and maintenance costs; User-friendly human-computer interaction: Supports multiple operation methods such as scanning codes, swiping cards, and account passwords; supports multiple payment channels such as WeChat, Alipay, and UnionPay; provides real-time visualization of charging information; and boasts a user operation success rate of ≥99.5%, enhancing the user experience. Highly scalable: The system adopts a modular design, with the hardware layer supporting the expansion of components such as sensors and power modules, and the software layer supporting OTA remote upgrades of algorithms and models, which can adapt to the technological development needs of future new energy vehicles.

[0062] Example of use To verify the effectiveness of the intelligent control system and method of this invention, the following four sets of experiments were designed. The experimental environment was a large parking lot in a city (5000㎡, including 200 parking spaces, and 5 Huidian Energy GA series mobile energy storage and charging robots deployed). The experimental equipment and parameters are as follows: Robot parameters: Energy storage capacity 120 kWh, maximum single-gun output power 120 kW, voltage range DC 150-1000V, maximum chassis speed 10 km / h, maximum gradeability 20%; Vehicles awaiting charging: Covering 10 mainstream Chinese standard models (BYD Han, Tesla Model 3, XPeng P7, NIO ES6, etc.) and 2 European standard models (BMW iX3, Mercedes-Benz EQC), with battery types including lithium iron phosphate and ternary lithium; Test equipment: high-precision voltage / current sensor (accuracy ±0.1%), temperature sensor (accuracy ±0.1℃), GPS locator (accuracy ±0.5m), power analyzer (accuracy ±0.2%), fault simulation generator.

[0063] Experiment 1: Charging Compatibility Test Experimental objective: To verify the charging compatibility of the present invention and test its adaptability to different vehicle models and battery types; Experimental steps: Twelve test vehicles were selected, and the battery type, voltage range, and charging protocol version of each vehicle were recorded. The robot receives a charging request and performs data collection, preprocessing, feature extraction, and compatibility judgment processes. Start charging and record whether charging was successful, charging power, and charging efficiency. Repeat the test 3 times and take the average value. The experimental data are shown in the table below: ; Results analysis: All 12 tested vehicles were successfully charged, with a compatibility of ≥0.83 and a charging efficiency of ≥94.7%, proving that the invention has extremely strong charging compatibility and can be adapted to different national / European standard vehicle models and different types of batteries.

[0064] Experiment 2: Battery State Estimation Accuracy Test Experimental objective: To verify the accuracy of SOC and SOH estimations; Experimental steps: Three vehicles with different battery conditions (new car, 1-year-old car, and 2-year-old car) were selected, and their true SOC and SOH were measured using a high-precision battery testing system (accuracy ±0.5%). The robot charges three vehicles, estimates SOC and SOH in real time, and records the error between the estimated value and the actual value. During the charging process, the ambient temperature was changed (-20℃, 0℃, 25℃, 50℃) to test the estimation accuracy at different temperatures; The experimental data are shown in the table below: ; Results analysis: The SOC estimation error is ≤ ±1.5%, the SOH estimation error is ≤ ±1.3%, and the accuracy remains high within the ambient temperature range of -20℃ to 50℃, proving that the battery state estimation model of the present invention is accurate and reliable.

[0065] Experiment 3: Path Planning and Scheduling Efficiency Test Experimental objective: To verify the energy consumption optimization effect and multi-robot cooperative scheduling efficiency of the improved A* path planning algorithm; Experimental steps: Five target parking spaces are set, located in different positions in the parking lot, and the straight-line distance between the robot's starting position and the target parking space is recorded; The traditional A algorithm and the improved A algorithm of this invention were used to plan the path, and the travel distance, travel time and energy consumption were recorded. Five robots were deployed to receive ten charging requests simultaneously. The multi-robot collaborative scheduling strategy of this invention was adopted to record the average waiting time of vehicles and the utilization rate of robots. Repeat the test 10 times and take the average value. The experimental data are shown in the table below: ; Multi-robot collaborative scheduling test data: ; Results analysis: The improved A algorithm reduces path energy consumption by more than 18% compared to the traditional A algorithm. Under multi-robot collaborative scheduling, the average waiting time of vehicles is only 4.35 minutes, and the robot utilization rate reaches 90%. This proves that the path planning and scheduling strategy of the present invention can effectively reduce energy consumption and improve scheduling efficiency.

[0066] Experiment 4: Fault Prediction and Operational Efficiency Test Experimental objective: To verify the accuracy of the LSTM fault prediction model and the effectiveness of proactive maintenance; Experimental steps: Five common faults (power module overheating, battery insulation failure, chassis motor failure, communication module failure, and charging gun poor contact) are injected into the robot using a fault simulation generator. The robot collects equipment status data in real time, predicts the probability of failure using an LSTM model, and records the accuracy of the failure prediction. Compared with traditional manual inspection, the proactive maintenance mode of this invention records fault response time and equipment downtime. The experimental data are shown in the table below: ; Equipment downtime comparison: ; Results analysis: The LSTM fault prediction model achieved an average accuracy of 92%, and the proactive maintenance response time was reduced by 59.7% compared to traditional manual inspection, equipment downtime was reduced by 62.4%, and equipment utilization was increased by 10.1%, proving that the fault prediction and maintenance strategy of this invention can effectively improve equipment reliability and maintenance efficiency.

[0067] Summary of the effects of the examples Through the above four sets of experiments, the intelligent control system and method for mobile energy storage and charging robots of the present invention have the following effects: Charging compatibility: Compatible with all tested Chinese and European standard models, with a compatibility rate of ≥83% and a charging efficiency of ≥94.7%; Battery state estimation: SOC estimation error ≤ ±1.5%, SOH estimation error ≤ ±1.3%, maintaining high accuracy in environments ranging from -20℃ to 50℃; Path planning and scheduling: Path energy consumption is reduced by more than 18% compared with traditional algorithms; the average waiting time of vehicles under multi-robot collaborative scheduling is ≤4.5min; and the robot utilization rate is ≥90%. Fault prediction and maintenance: Fault prediction accuracy ≥92%, maintenance response time reduced by 59.7%, equipment downtime reduced by 62.4%, and equipment utilization rate increased to over 95%.

[0068] Experimental results fully demonstrate that the present invention solves the problems of poor compatibility, crude battery management, unintelligent scheduling, and untimely operation and maintenance in the prior art, significantly improves the intelligence level, operational efficiency, and safety of mobile energy storage and charging robots, and achieves the expected invention objectives.

[0069] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. An intelligent control method for a mobile energy storage and charging robot, characterized in that, Using a mobile energy storage and charging robot as the main implementer, the process includes the following steps: Data acquisition steps: The robot collects multi-source initial data in real time through its onboard sensor group, communication module and interaction interface. The multi-source initial data includes the robot's own status data, the charging demand data of the electric vehicle to be charged, external environment data and scheduling instruction data issued by the cloud platform. Data preprocessing steps: The robot performs noise reduction, standardization, anomaly detection, and data completion on the multi-source initial data to obtain standardized data; Feature extraction step: Based on the standardized data, the robot extracts multi-dimensional feature data through a preset feature extraction algorithm. The multi-dimensional feature data includes battery status features, charging demand adaptation features, environmental adaptation features, and location scheduling features. Decision-making steps: The robot inputs the multi-dimensional feature data into the preset decision model, and performs charging compatibility judgment, battery safety status judgment, charging strategy matching judgment, path planning judgment and operation and maintenance early warning judgment respectively, and outputs the corresponding decision instructions; Execution control steps: The robot executes corresponding control operations according to the decision instructions, including movement control, dynamic adjustment of charging parameters, battery protection control, cloud interaction control, and human-machine interaction control; Feedback optimization steps: The robot collects the execution feedback data of the control operation, and combines it with the optimization instructions of the cloud platform to iteratively optimize the feature extraction algorithm, decision model and control parameters, and update the preset algorithm and model parameters.

2. The intelligent control method according to claim 1, characterized in that, The robot's own state data includes: Energy storage system data: individual cell voltage, total voltage, charge / discharge current, individual cell temperature, total temperature, initial SOC, initial SOH, battery pack insulation resistance, coolant flow rate and temperature; Power system data: power module input voltage, output voltage, output current, module temperature, operating efficiency, and cooling fan speed; Chassis drive module data: rated power, peak power, rated torque, peak torque, speed, current, voltage, chassis attitude angle, driving speed, driving distance, steering angle, braking status, obstacle crossing status of chassis motor; Equipment status data: monitoring the operating status of the core board, the signal strength and connection status of the communication module, the operation records of the human-machine interaction module, and payment verification data.

3. The intelligent control method according to claim 1, characterized in that, The vehicle charging demand data includes: battery type, rated battery capacity, voltage range, charging power demand, initial SOC, target SOC, charging protocol type, vehicle location coordinates, and vehicle identification information of the vehicle to be charged. External environmental data includes: ambient temperature, ambient humidity, altitude, atmospheric pressure, precipitation status, and obstacle information; the scheduling instruction data issued by the cloud platform includes: scheduling priority, peak and valley time period identifiers, charging price information, operation and maintenance task instructions, and algorithm update instructions.

4. The intelligent control method according to claim 1, characterized in that, The data preprocessing steps specifically include: Noise Reduction: The Kalman filter algorithm is used to reduce noise in the battery cell voltage, current, and temperature data. The filtering equation is: Prediction equation: ; Filter gain equation: ; Update equation: ; in, The data at time k is the filtered data. Here is the state transition matrix. For the input matrix, Input data for time k-1. For filter gain, To predict the error covariance, For the observation matrix, To observe the noise covariance, The data is the observation data at time k; Standardization: Min-max standardization is used to normalize data such as voltage range and power demand. The formula is: ; in, For standardized data, This is the original data. The minimum value of the data. The maximum value of the data; Anomaly detection: The 3σ criterion is used to remove outliers from the data. When the data meets the criteria... When it is determined to be an outlier, among which The mean of the data. The standard deviation of the data; Data completion: Missing data is completed using linear interpolation. The formula is as follows: ； in, For missing data, , These are the valid data before and after the missing data, respectively. , , These represent the times when the corresponding data was collected.

5. The intelligent control method according to claim 1, characterized in that, The feature extraction steps specifically include: Battery state feature extraction: SOC characteristics: SOC is estimated based on the ampere-hour integral method combined with the extended Kalman filter (EKF), and the formula is as follows: ； in, For the initial SOC, For the battery's rated capacity, This is the charging and discharging current. For EKF correction terms; SOH characteristics: SOH is estimated based on a nonlinear regression model, using the following formula: ； in, For the current battery capacity, This refers to the battery's internal resistance. The number of loops. This represents the average battery temperature. ~ , For model parameters, This is the error term; Battery consistency characteristics: Calculating the standard deviation of individual cell voltages and temperature standard deviation Consistency coefficient ,in This refers to the battery's nominal voltage. Charging demand adaptation feature extraction: compatibility ; in, For voltage range matching, For power demand matching, For charging protocol compatibility, , , These are the weighting coefficients. ; Environmental adaptation feature extraction: Environmental adaptability coefficient ; in, For temperature adaptation coefficient, Humidity adaptation factor, This refers to the altitude adaptation factor. , , These are the weighting coefficients; Location scheduling feature extraction: Path length feature: Calculate the straight-line distance based on the current location and the coordinates of the target parking space. ; Path cost characteristics: Calculate path cost by considering slope and obstacle crossing conditions. ,in For distance, The slope coefficient is... The obstacle crossing coefficient, For energy consumption, , , , These are the weighting coefficients.

6. The intelligent control method according to claim 5, characterized in that, The decision-making steps specifically include: Charging compatibility assessment: When the compatibility is... When it is determined to be compatible, a "Charging Allowed" command is output; when When it is determined to be partially compatible, a "reduce power charging" command is output; when If the condition is not met, a "refuse charging" command will be output. Battery safety status assessment: Overvoltage detection: When the voltage of a single battery cell... When this occurs, an overvoltage protection command is output; Over-temperature judgment: When the battery temperature When, output the liquid cooling start command; when When the time comes, output a stop charging command; Insulation judgment: When the insulation resistance When this occurs, output the "insulation fault" command; Charging strategy matching judgment: Constant current mode: when and At that time, a constant current charging command is output, and the charging current is... ; Constant pressure mode: when At that time, a constant voltage charging command is output, and the charging voltage is... ; Trickle mode: when At that time, a trickle charging command is output, and the charging current is... ; Path planning decision: Based on the improved A* algorithm, using path cost With the goal of minimizing the minimum, plan the optimal driving path and output the path execution command. The path planning formula is: ,in The cost from the starting node to the current node. This is the heuristic function from the current node to the target node; Operation and maintenance early warning judgment: predicting the probability of failure based on LSTM neural network model ,when When, output operation and maintenance early warning commands; when When this happens, an emergency stop command is output.

7. The intelligent control method according to claim 6, characterized in that, The execution control steps specifically include: Motion control: Based on path planning instructions, control the chassis drive module to perform driving operations, including front wheel steering angle. driving speed Dynamically adjust based on path complexity, straight line segment Turning section Obstacle crossing section ; Dynamic adjustment of charging parameters: Based on charging strategy instructions, the output voltage and current are adjusted through the power module. Voltage adjustment formula. ,in For the vehicle's required voltage, For optimal temperature, , This is the adjustment coefficient; Battery protection control: Based on safety status commands, the corresponding protection mechanism is activated. Overvoltage occurs, the charging circuit is cut off; overtemperature occurs, the liquid cooling system is activated, and the coolant flow rate is adjusted. Disconnect the high-voltage box in case of an insulation fault; Cloud-based interactive control: The charging data and device status data are uploaded to the cloud platform via the 4G / WIFI communication module, and the scheduling instructions and algorithm update instructions from the cloud platform are received. Data transmission adopts an encrypted protocol. Human-computer interaction control: The touch screen displays information such as charging progress, SOC, and charging cost. It receives user's QR code scanning, account password, and card payment instructions. Charging starts after payment verification and a charging report is generated and displayed after payment is completed.

8. The intelligent control method according to claim 6, characterized in that, The feedback optimization steps specifically include: Collect feedback data: including charging efficiency SOC estimation error SOH estimation error Charging time deviation Number of equipment failures; Algorithm optimization: Adjusting the parameters of the feature extraction algorithm based on gradient descent, and optimizing the process noise covariance of EKF. Observation noise covariance The formula is: in, For parameters to be optimized, For learning rate, Let it be the error loss function; Model update: Input feedback data into the decision model for retraining, update model parameters, and adjust fit weights. The weight matrix of the fault prediction model; Parameter storage: The optimized algorithm parameters and model parameters are stored in the flash memory of the monitoring core board and automatically loaded on the next startup.

9. An intelligent control system for a mobile energy storage and charging robot, characterized in that, The intelligent control method according to any one of claims 1-8, wherein the system comprises a hardware layer and a software layer, and the hardware layer and the software layer are connected via a bus communication connection: Hardware layer: Monitoring core board: adopts automotive-grade STM32H743VI chip, integrating CPU, memory, flash memory, CAN bus interface, RS232 serial port, and Ethernet interface to realize system monitoring, coordination control, data processing and cloud interaction functions; Energy storage system: includes lithium iron phosphate battery pack, high voltage box, and BMS. The BMS uses TI's BQ79616 chip to realize battery status monitoring, equalization control and safety protection. Power system: includes DC / DC power module, filter unit, and metering unit, wherein the power module adopts forced air cooling; Chassis drive module: includes Ackermann chassis, drive motor, steer-by-wire EPS system, EHB electro-hydraulic braking system, and front and rear coil spring suspension, to realize robot movement control; Communication module: integrates 4G module, WIFI module, and Bluetooth module to enable communication with cloud platform, vehicle to be charged and human-machine interaction module; Sensor group: includes voltage sensor, current sensor, temperature sensor, GPS positioning module, lidar, and ultrasonic sensor; Human-computer interaction module: including touch screen, QR code scanning module, card swiping module, status indicator light, and advertising screen; Protection module: includes overvoltage protection circuit, overcurrent protection circuit, overtemperature protection circuit, insulation detection circuit, and emergency stop button; Software layer: Data acquisition module: drives the sensor group, BMS, and power module to acquire data from multiple sources, and realizes data caching and synchronization; Data preprocessing module: integrates Kalman filtering, min-max normalization, 3σ anomaly detection, and linear interpolation functions to process the raw data; Feature extraction module: Extracts battery status features, charging demand adaptation features, environmental adaptation features, and location scheduling features; Decision control module: includes compatibility judgment unit, security judgment unit, policy matching unit, path planning unit, and operation and maintenance early warning unit, and outputs decision instructions; Execution control module: Drives the chassis drive module, power module, BMS, and communication module to perform control operations; Cloud interaction module: Enables data uploading, command reception, algorithm updates, and remote operation and maintenance with the cloud platform; Operation and maintenance optimization module: Based on feedback data, optimize algorithms and model parameters to generate operation and maintenance reports.

10. The intelligent control system according to claim 9, further comprising a cloud platform interaction unit, the cloud platform comprising: Data storage module: Stores robot status data, charging records, user data, and operation and maintenance data, using a distributed database; Scheduling Management Module: Implements multi-robot collaborative scheduling based on a greedy algorithm. The scheduling objective function is: ； in The charging revenue for the i-th robot, For energy consumption costs, As scheduling priority weights; Operation and maintenance management module: Generates inspection plans based on big data analysis, predicts equipment failures, and outputs operation and maintenance instructions; User management module: Manages user accounts, payment information, and charging records, and supports WeChat, Alipay, and UnionPay payments; Algorithm update module: Sends algorithm update packages to the robot, supporting OTA remote upgrades.

Images