Charging control method for battery pack of narrow-body self-walking working platform

Abstract

The invention discloses a narrow-body self-walking operation platform battery pack charging control method, and belongs to the technical field of battery management. Operating parameters of a lithium battery pack and working condition parameters of a working platform are collected in real time, corresponding charging stages are selected according to charge states and working condition types, and the charging process is divided into five stages including pre-charging, quick charging, optimized charging, equalizing charging and floating charging maintenance. A three-dimensional temperature field model is established based on multi-point temperature data, and a comprehensive temperature compensation coefficient is calculated to accurately adjust the charging current. The state of charge and the state of health of the battery are jointly estimated by adopting extended Kalman filtering and a long-short term memory neural network, and the estimation precision is improved through Bayesian fusion. The inclination angle and vibration acceleration of the platform are monitored in real time, and charging is adjusted or stopped according to safety conditions. The charging time is remarkably shortened, the charging efficiency is improved, the battery temperature rise is reduced, the battery cycle life is prolonged, and meanwhile the energy utilization efficiency and safety of the operation platform are improved.

Description

Technical Field

[0001] This invention belongs to the field of battery management technology, specifically relating to a method for controlling the charging of battery packs in a narrow-body self-propelled work platform. Background Technology

[0002] Existing lithium battery charging technologies primarily employ the traditional constant-current, constant-voltage charging method. This method charges the battery with a fixed current during the constant-current phase, then switches to constant-voltage charging once the battery voltage reaches a set value, continuing until the charging current decays to a cutoff value. While this method is simple in structure and easy to implement, it suffers from fixed charging parameters that cannot be dynamically adjusted according to actual working conditions. For mobile work equipment like aerial work platforms, operations involve various states, including stationary standby, mobile movement, and boom operation, each with significantly different power requirements. Traditional fixed charging modes may consume excessive engine power during heavy-load operations, leading to insufficient operating power, while during light-load or standby conditions, the remaining power cannot be fully utilized for rapid charging, resulting in low charging efficiency and impacting operational continuity. Existing charging technologies typically employ single-stage or two-stage charging strategies, failing to fully consider the differences in charging characteristics of lithium batteries under different states of charge.

[0003] Batteries can accept larger charging currents when at a low state of charge (SOC), but their charging acceptance decreases when at a high SOC. Using the same charging current at this high SOC can easily lead to overcharging and excessive battery temperature rise. Furthermore, existing methods are relatively crude in their charging phase divisions and current settings, making it difficult to achieve a good balance between charging speed, charging efficiency, and battery life. Temperature is a key factor affecting the charging performance and safety of lithium batteries, but current technologies mostly use single-point or a few temperature measurement points to monitor battery temperature, failing to accurately reflect the internal temperature distribution of the battery pack. During charging, the battery pack generates temperature gradients due to uneven current distribution and differences in heat dissipation conditions, potentially leading to overheating in localized areas. Single-point temperature measurement can easily overlook localized temperature anomalies, resulting in improper charging parameter settings that affect charging efficiency and pose safety hazards. Especially in extreme temperature environments, inaccurate temperature compensation can cause decreased charging acceptance at low temperatures and accelerated battery aging at high temperatures. Accurate estimation of the battery's SOC and state of health is fundamental to charging control. Current technologies mainly use the ampere-hour integral method or model-based Kalman filtering algorithms to estimate the SOC. While the ampere-hour integration method is simple and intuitive, it suffers from cumulative error, leading to decreased estimation accuracy over prolonged use. The Kalman filter algorithm relies on accurate battery models, but battery characteristics change with temperature and aging, making real-time model parameter correction difficult and resulting in low estimation accuracy. For health status estimation, existing methods largely depend on offline testing, failing to acquire real-time battery health information during charging. Charging parameters cannot adaptively adjust based on battery aging, accelerating performance degradation. For work platforms equipped with dual power sources, current technologies lack intelligent management of power source switching and power distribution. Furthermore, the lack of energy recovery mechanisms means that mechanical energy during boom descent and vehicle deceleration is not effectively recovered, further reducing energy efficiency. Existing charging technologies also suffer from the inability to continuously optimize charging strategies. Charging parameters are typically fixed at the factory and cannot be adjusted based on actual usage and battery aging. With increased usage time and changing operating conditions, initially set charging parameters become inapplicable, leading to decreased charging performance. Although some high-end equipment is equipped with data logging capabilities, the data is not effectively utilized, lacking big data-driven charging strategy optimization and predictive maintenance functions.

[0004] In summary, there is an urgent need to develop a battery pack charging control method for narrow-body self-propelled work platforms to solve the aforementioned technical problems, improve charging efficiency and battery life, reduce operating costs, and ensure operational safety. Summary of the Invention

[0005] To address the problems existing in the background art, the present invention provides a battery pack charging control method for a narrow-body self-propelled work platform, characterized by comprising the following steps: S1: Collect lithium battery pack operating parameters and operating platform parameters; S2: Construct operating condition feature vectors to identify the current operating condition type; S3: Select the charging stage based on the operating condition type and SOC value, and determine the reference charging current; S4: Establish a three-dimensional temperature field model based on multi-point temperature data, calculate the comprehensive temperature compensation coefficient, and adjust the charging current; S5: Use extended Kalman filtering and long short-time memory neural network to jointly estimate SOC and SOH, and correct the charging parameters; S6: Predict future power demand, switch the power source between the diesel engine and the electric pump, and dynamically allocate charging power; S7: Detect the platform tilt angle and vibration acceleration, and adjust or stop charging; S8: Upload charging and discharging data to the cloud server, receive the optimized charging strategy parameters, and update the local algorithm.

[0006] Furthermore, S1 includes the following steps: S11: The voltage of each individual cell in the battery pack is acquired using a voltage acquisition module at a sampling frequency of 100Hz to 1000Hz, with a voltage acquisition accuracy of ±5mV to ±10mV; S12: The charging current or discharging current is acquired using a current sensor connected in series in the main charging and discharging circuit at a sampling frequency of 500Hz to 2000Hz, with a current sensor measurement accuracy of ±0.3% to ±0.8%; S13: The temperature at multiple points within the battery pack is acquired using 12 to 20 temperature sensors, which are distributed at the positive terminal, negative terminal, the middle of the battery pack, and the outer casing. S14: The walking speed is measured by the walking speed sensor installed on the walking wheels, the steering angle is measured by the steering angle displacement sensor installed on the steering mechanism and the steering angle velocity is calculated, and the hydraulic system power is calculated by the pressure sensor and flow sensor installed on the hydraulic system oil pipes; S15: The tilt angle and vibration acceleration of the platform in the X, Y and Z directions are measured by the six-axis attitude sensor installed at the center of the walking platform; S16: All collected parameters are transmitted to the main controller through the CAN bus, and the main controller performs validity verification and filtering processing on the data.

[0007] Furthermore, S2 includes the following steps: S21: Constructing a seven-dimensional feature vector for operating conditions: ;in, Walking speed; This refers to the steering angular velocity; Power of the hydraulic system; The platform tilt angle; It is the vibration acceleration; This is the current driving the motor; S22: Normalize the seven-dimensional working condition feature vector using the following formula: [Formula omitted for brevity] ;in, is the normalized value of the i-th feature; This represents the actual measured value of the i-th feature; The minimum value of the i-th feature; S23: Input the normalized feature vector into the pre-trained fuzzy neural network, which includes 7 input nodes, 15 to 25 hidden layer nodes and 7 output layer nodes; S24: Calculate the output value of each node in the output layer, select the working condition type corresponding to the node with the largest output value as the recognition result, and identify it as one of the following working conditions: stationary standby working condition, low-speed walking working condition, high-speed walking working condition, steering maneuvering working condition, boom light load action working condition, boom heavy load action working condition or mixed operation working condition.

[0008] Furthermore, S3 includes the following steps: S31: Determine the current SOC value range. When SOC is less than 0.20, select the pre-charging stage; when SOC is between 0.20 and 0.60, select the fast charging stage; when SOC is between 0.60 and 0.85, select the optimized charging stage; when SOC is between 0.85 and 0.95, select the equalization charging stage; when SOC is greater than or equal to 0.95, select the float charging maintenance stage. S32: Determine the reference charging current based on the selected charging stage and the current operating condition. During the pre-charging phase, settings are configured for stationary standby, low-speed travel, and high-speed travel conditions. The setting is 0.15C to 0.25C, for steering maneuvering and boom operation conditions. The voltage range is 0.03C to 0.07C; during the fast charging phase, the voltage is set under standby conditions. The current is set to 0.45C to 0.55C. During the walking operation, pulse charging is used and the average current is calculated. During the boom operation, the current is set... The charging rate is 0.25C to 0.35C; S33: During the optimized charging phase, according to the formula... Calculate the reference charging current, where The maximum charging current is set between 0.4C and 0.5C; during the equalization charging phase, the voltage difference between individual cells is monitored. According to the formula Calculate the reference charging current, where This is the difference between the highest and lowest voltage of a single battery cell; during the float charging maintenance phase, constant voltage charging is used, with the charging voltage set to 54.4V to 54.8V; S34: Determine the operating condition adjustment coefficient according to the operating condition type. In standby mode Low-speed travel conditions Up to 0.95, under high-speed driving conditions Up to 0.85, under steering maneuvering conditions Up to 0.80, under light load operating conditions of the boom Up to 0.90, under heavy-load boom operation conditions Up to 0.70, under mixed operating conditions Up to 0.75; S35: Calculate the adjusted charging current ;in This is the adjusted charging current.

[0009] Furthermore, S4 includes the following steps: S41: Read the real-time temperature data from all temperature sensors to obtain the temperature vector. ,in S42: Determine the position weighting coefficient based on the position of each temperature sensor. The weighting coefficient at the positive and negative terminals is 0.15 to 0.20, the weighting coefficient at the middle of the battery pack is 0.10 to 0.15, and the weighting coefficient at the surface of the battery pack casing is 0.05 to 0.10. S43: Calculate the single-point temperature compensation factor for the temperature at each measuring point. :when When the temperature is below -10℃ Up to 0.6, when Between -10℃ and 0℃ to ;when Between 0℃ and 20℃ to ;when Between 20℃ and 35℃ Up to 1.05; when Between 35℃ and 50℃ to ;when When the temperature is above 50℃ Up to 0.5; S44: Calculate the comprehensive temperature compensation coefficient using a weighted average method: ;in This is the comprehensive temperature compensation coefficient; The position weight coefficient of the i-th temperature measurement point; Σ represents the summation over all temperature measurement points; S45: Calculate the maximum temperature gradient within the battery pack: ;in The maximum temperature gradient; The absolute value of the temperature difference between any two temperature measurement points; The distance between two temperature measuring points; when When it exceeds 5℃ / dm, Further reduced to to Simultaneously activate the air-cooling system; S46: Correct the charging current based on the comprehensive temperature compensation coefficient, and calculate the temperature-compensated charging current. ;in This is the charging current after temperature compensation.

[0010] Furthermore, S5 includes the following steps: S51: Establish a second-order RC equivalent circuit model and define the state vector: ;in Let k be the state of charge at time k; and These represent the voltages of the two RC branches; T denotes matrix transpose; S52: Execute the extended Kalman filter algorithm to calculate the prior state estimate: ;in For prior state estimation; A represents the posterior state estimate from the previous time step; A is the state transition matrix; B is the input matrix. S53: Calculate the Kalman gain and update the posterior state estimate, extracting the charge / discharge current from the posterior state estimate. As the SOC estimate for the extended Kalman filter algorithm; S54: Construct a feature vector containing 30 consecutive sampling points, the feature vector including SOC, voltage, current, temperature, internal resistance, voltage change after resting, and cumulative loop count. Input the feature vector into a trained long short-term memory neural network, the neural network including a first LSTM layer, a Dropout layer, a second LSTM layer, and a fully connected layer, and output... and S55: Calculate the confidence variance of the extended Kalman filter and neural network estimates, and calculate the weight coefficients based on the confidence variance. and ;in The confidence variance estimated by extended Kalman filtering; S56: Calculate the final SOC and SOH values ​​using Bayesian fusion. ;in This is a correction factor, ranging from -0.8 to -1.2; This represents the relative change in internal resistance; S57: Corrects the charging parameters based on the final SOH value, when... When the current is less than 0.90, the maximum charging current will be reduced to 0.35C to 0.45C. When the value is less than 0.85, the maximum charging current is further reduced to 0.30C to 0.40C. When the voltage is less than 0.80, the charging cutoff voltage will be reduced by 0.03V to 0.08V.

[0011] Furthermore, S6 includes the following steps: S61: Collect historical power data from the past 5 to 10 minutes, establish a power series, and use a sliding window time series analysis method to predict the power demand for the next 30 to 60 seconds. The prediction formula is: ;in Δt represents the predicted power; Δt represents the prediction duration. Historical data weighting coefficients; Historical power values; S62: Determine the power source switching conditions. When the load power is greater than 0.80 to 0.90 times the maximum power of the electric pump, or the SOC is less than 0.25 to 0.35, or the predicted power is greater than 0.85 to 0.95 times the rated power of the electric pump, switch from the electric pump to the diesel engine. When the load power is less than 0.55 to 0.65 times the output power of the diesel engine, and the SOC is less than 0.85 to 0.95, and the duration is greater than 8s to 12s, switch from the diesel engine to the charging mode. S63: In the charging mode, calculate the remaining power available for charging. ;in The remaining power available for charging; This refers to the output power of the diesel engine. For load power; S64: Calculate the charging power, which is the reserved power determined based on the predicted power. ;in This represents the actual charging power; min indicates taking the minimum value. This is the maximum charging power; For charging efficiency, the value ranges from 0.88 to 0.95; This is the power adjustment factor based on SOC; S65: Temperature-based power adjustment factor; Calculate charging current based on charging power and battery voltage. ;in This refers to the charging current calculated based on power allocation. The current voltage of the battery pack; S66: When the boom is lowered or the vehicle is decelerated, detect whether the drive motor is in the generating state. When it is in the generating state and the energy recovery conditions are met, start the energy recovery mode, calculate the energy recovery current and limit it to the range of 0.3C to 0.6C.

[0012] Furthermore, S7 includes the following steps: S71: Read the platform's forward / backward tilt angle α and left / right tilt angle β from the six-axis attitude sensor, and calculate the overall tilt angle. ;in S72: Adjust the charging current according to the overall tilt angle; when The charging current remains constant when the temperature is less than 3°; when The charging current is adjusted when the temperature is between 3° and 5°. to ;when The charging current is adjusted when the temperature is between 5° and 8°. to ;when Charging stops and an audible and visual alarm is triggered when the angle is greater than or equal to 8°; S73: Reads three-axis acceleration data from the six-axis attitude sensor. , and Calculate the overall vibration acceleration ,in For comprehensive vibration acceleration; S74: Adjust the charging current according to the comprehensive vibration acceleration: when If the amount exceeds 0.4g to 0.6g and lasts for more than 2s to 4s, reduce the charging current to [a lower value]. to ;when Charging is immediately paused when the amount exceeds 0.8g to 1.2g, and charging resumes after a 3-7 second delay once the vibration returns to normal; S75: Monitors voltage fluctuations in the charging circuit and calculates the voltage fluctuation rate. ;in Voltage fluctuation rate; The maximum voltage within the sampling window; Minimum voltage; For average voltage; when If the current exceeds 0.04 to 0.06, it is determined that the connector may be loose, the charging current is reduced to 0.10C to 0.20C and an alarm is triggered; S76: Select the smaller value between the charging current after tilt adjustment and the charging current after vibration adjustment as the final safe charging current and send it to the charging controller for execution.

[0013] Furthermore, S8 includes the following steps: S81: After each operation, read charge / discharge curve data, historical temperature data, estimated SOC and SOH values, operating condition type statistics, abnormal event records, and environmental parameters from the data storage module; S82: Package the data into a JSON or XML data packet, which includes vehicle identification code, operation date, battery data, operating condition data, abnormal event data, and environmental data fields; S83: Upload the data packet to the cloud server using HTTPS protocol via a 4G / 5G wireless communication module or WiFi module; S84: The cloud server receives data from multiple operation platforms and uses a big data processing framework to perform cluster analysis of similar operating conditions, evaluation of charging strategy effectiveness, training of battery life prediction models, and statistical analysis of optimal charging parameters; S85 S86: The cloud server generates optimized charging strategy parameters based on big data analysis results. These parameters include current setpoints for each charging stage, charging cutoff voltage, equalization charging time, temperature compensation coefficient matrix, and operating condition adjustment coefficient. S87: The cloud server packages the optimized charging strategy parameters into a configuration file and pushes it to the work platform. S88: The main controller of the work platform receives the configuration file, performs integrity verification and version compatibility checks, and then updates the parameters of the charging stage selection unit, temperature compensation processing unit, and energy management unit. S89: The cloud server predicts the remaining battery life (RUL) and failure probability based on battery aging trends. When RUL is less than a preset threshold or failure probability is greater than a preset threshold, it generates an early warning message and pushes it to the work platform and management platform.

[0014] Furthermore, in step S32, during the walking operation in the fast charging phase, a pulse charging mode is adopted, and the duty cycle of the pulse charging is... ;in Duty cycle; Current walking power; Maximum walking power; pulse period is 2s to 5s, current is 0.5C during charging pulse, current is 0.05C to 0.10C during discharging pulse, and charging pulse duration is... The discharge pulse duration is ;in The pulse period; This refers to charging time; This refers to the discharge time.

[0015] The beneficial effects achieved by this invention are as follows: First, this invention employs a technical solution combining an adaptive multi-stage charging strategy with operating condition adjustment. The charging process is divided into five stages: pre-charging, fast charging, optimized charging, equalization charging, and float charging maintenance. Each stage uses a different charging current setting based on the SOC value range, while an operating condition adjustment coefficient is introduced to correct the reference charging current. Under low-load conditions, the charging current is increased to shorten charging time, while under high-load conditions, the charging current is reduced to ensure operational power requirements, achieving an intelligent balance between charging speed and operational needs. Compared to traditional constant-current, constant-voltage charging methods, this multi-stage adaptive strategy better adapts to the changing charging characteristics of lithium batteries in different SOC ranges, improving charging acceptance while avoiding overcharging risks. Under mixed operating conditions, it completely eliminates the problem of insufficient power, effectively shortening charging time and providing reliable energy assurance for the efficient operation of the work platform.

[0016] Secondly, this invention achieves adaptive response of the charging strategy to the working conditions through multi-dimensional parameter acquisition and working condition identification. It constructs a seven-dimensional working condition feature vector and inputs it into a pre-trained fuzzy neural network for working condition identification. This charging control method based on real-time working conditions breaks through the limitations of the traditional fixed charging mode, enabling the charging current to be dynamically adjusted according to the working state. It fully utilizes the engine power for rapid charging when stationary and idles, and reduces the charging current to ensure priority power supply during heavy-load operations. It significantly shortens the charging time across the entire temperature range, improves charging efficiency, reduces battery temperature rise, and eliminates the frequent power shortage alarms in traditional methods, thereby improving the continuity and reliability of the working platform.

[0017] Third, this invention establishes a three-dimensional temperature field model and employs a joint estimation technique using extended Kalman filtering and long short-term memory neural networks, improving the precision of charging control and the accuracy of battery state estimation. It calculates a comprehensive temperature compensation coefficient based on the position weights of each temperature measurement point and the temperature compensation factor, and monitors the maximum temperature gradient to prevent local overheating. Compared to single-point temperature measurement methods, this approach more comprehensively reflects the true thermal state of the battery pack, avoiding improper charging parameter settings due to local temperature anomalies. Regarding SOC and SOH estimation, the system integrates a model-based extended Kalman filtering algorithm and a data-based long short-term memory neural network algorithm, achieving complementary advantages through Bayesian fusion. It also adaptively adjusts charging parameters based on the SOH value to adapt to battery aging. This significantly reduces SOC estimation errors across the entire temperature range, and precise temperature compensation effectively reduces charging temperature rise. The strategy of adaptively adjusting parameters based on SOH significantly extends battery cycle life, slows capacity decay and internal resistance growth, and improves the consistency and long-term reliability of the battery pack.

[0018] Fourth, this invention uses a sliding window time series analysis method to predict future power demand and makes decisions on switching power sources between the diesel engine and the electric pump in advance based on the prediction results, avoiding the problems of frequent power source switching and insufficient power supply. In charging mode, the system accurately calculates the remaining power of the diesel engine and dynamically allocates charging power. When the boom lowers or the vehicle decelerates, the energy recovery mode is activated to convert mechanical energy into electrical energy to recharge the battery pack. Compared with passive response control, this more rationally coordinates the energy allocation between charging and operation, reduces unnecessary engine running time and idling conditions, significantly reduces the cumulative running time and fuel consumption of the diesel engine, improves charging and discharging efficiency and overall energy utilization efficiency, and the energy recovery function further enhances the energy utilization effect. While reducing operating costs, it also reduces carbon emissions, which meets the requirements of energy conservation and environmental protection.

[0019] Fifth, this invention monitors the tilt angle and vibration acceleration of the platform in real time. Based on the magnitude of the combined tilt angle and vibration acceleration, it adjusts the charging current or stops charging and triggers an alarm in stages. It monitors voltage fluctuations in the charging circuit to identify loose connectors and responds promptly to tilting and vibration conditions caused by uneven ground or boom movements. This avoids safety hazards such as battery internal imbalance, overheating, and connection failures caused by continued charging under abnormal conditions. The cloud-based collaborative system uploads charging and discharging data, temperature history, operating condition statistics, and abnormal events to the server for big data analysis. Through clustering of similar operating conditions, evaluation of charging strategy effectiveness, and training of battery life prediction models, it generates optimized parameters and pushes them to the operating platform, enabling remote upgrades and predictive maintenance of the charging algorithm. Multiple safety protection functions effectively prevent safety accidents under extreme conditions such as tilting and vibration, significantly reducing the number of safety incidents and improving the system's intelligence level and overall lifecycle economy. Attached Figure Description

[0020] Figure 1 This is a comparison chart of charging time, maximum temperature rise, charging efficiency, and SOC estimation error at different temperatures in Experiment 1. Figure 2 The chart shows the adaptive adjustment strategy of charging current under different working conditions in Experiment 2, and the performance comparison chart of 50 cycles of mixed operation. Figure 3 The cycle life test results of Experiment 3 include a comparison of capacity retention, internal resistance growth rate, overall performance after 900 cycles, and predicted lifetime. Figure 4 This is a comparison chart of energy utilization efficiency in Experiment 4, including engine running time, diesel consumption, charging and discharging energy, and overall efficiency. Figure 5 This is the safety protection function verification diagram for Experiment 5, including tilt protection, vibration protection, and safety event statistics. Figure 6 This is a flowchart of steps S1-S3 of the battery pack charging control method for the narrow-body self-propelled work platform of the present invention; Figure 7 This is a flowchart of steps S4-S6 of the battery pack charging control method for the narrow-body self-propelled work platform of the present invention; Figure 8 This is a flowchart of steps S7-S8 of the battery pack charging control method for the narrow-body self-propelled work platform of the present invention; Figure 9 This is a schematic diagram of the structural composition of the narrow-body self-propelled work platform in Example 1. Figure 1 ; Figure 10 This is a schematic diagram of the structural composition of the narrow-body self-propelled work platform in Example 1. Figure 2 .

[0021] Numbering on the map: 1. Walking platform; 11. Wheeled four-wheel drive chassis; 111. Steering mechanism; 12. Diesel engine; 13. Diesel tank; 14. Electric pump; 15. Battery pack; 2. Working device; 21. Turntable; 22. Main telescopic boom; 23. Hydraulic mechanism I; 24. Insulated boom; 25. Hydraulic mechanism II; 26. Slewing mechanism; 27. Working bucket; 3. Control system; 31. Control box; 32. Remote control. Detailed Implementation

[0022] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Reference Figures 1-10 The battery pack charging control method for the narrow-body self-propelled work platform provided by this invention achieves intelligent charging management of the 48V lithium battery pack through multi-dimensional parameter acquisition, intelligent working condition identification, adaptive multi-stage charging, precise temperature compensation, joint estimation of SOC / SOH, predictive power allocation, safety protection, and cloud-based collaborative optimization.

[0024] Reference Figures 6 to 8 In step S1, the system collects the operating parameters of the lithium battery pack and the operating parameters of the work platform in real time, providing a data foundation for subsequent charging control. This step acquires battery status information and platform operating status information through various sensors and acquisition modules, ensuring that the charging control strategy can respond to changes in actual operating conditions.

[0025] In step S11, the voltage of each individual cell in the battery pack 15 is acquired by a voltage acquisition module at a sampling frequency of 100Hz to 1000Hz, with a voltage acquisition accuracy of ±5mV to ±10mV. The voltage acquisition module employs a high-precision analog-to-digital converter (ADC), which converts analog voltage signals into digital signals. A 16-bit or 24-bit resolution ADC chip is preferred to ensure voltage measurement accuracy. Multiple acquisition channels of the voltage acquisition module are connected to the positive and negative terminals of each individual cell in the battery pack. For a 48V lithium battery system, a 13-cell configuration is typically used, requiring 14 acquisition channels. The selection of the sampling frequency needs to balance data accuracy and processor load; a sampling frequency of 500Hz to 1000Hz is preferred, as it captures dynamic voltage changes during charging without causing data processing overload. High sampling frequency and high-precision voltage acquisition can promptly detect abnormal individual cell voltages, providing accurate data for subsequent equalization charging and safety protection.

[0026] In step S12, the charging or discharging current is acquired by a current sensor connected in series in the main charging / discharging circuit at a sampling frequency of 500Hz to 2000Hz. The current sensor measurement accuracy is ±0.3% to ±0.8%. Hall effect current sensors or shunt current sensors are preferred. Hall effect sensors, based on the principle of magnetic field induction, have the advantages of non-contact measurement, fast response speed, and good linearity, making them suitable for high-current measurement. Shunt current sensors calculate the current by measuring the voltage drop across a precision resistor, featuring high accuracy and good temperature stability. The sampling frequency of the current sensor needs to be higher than the voltage sampling frequency, preferably 1000Hz to 2000Hz, to accurately capture the pulse changes and transient response of the charging current. The current data is transmitted to the main controller via a CAN bus. CAN stands for Controller Area Network, a serial communication protocol used in automotive and industrial control, characterized by strong anti-interference capabilities and good real-time performance. Accurate current measurement is crucial for achieving charging power control and SOC estimation, directly affecting the execution effect of the charging strategy.

[0027] In step S13, the temperature of multiple points within the battery pack 15 is collected using 12 to 20 temperature sensors. These sensors are distributed at the positive terminal, negative terminal, the middle of the battery pack, and the outer surface. NTC thermistors or digital temperature sensors such as the DS18B20 are preferred. NTC stands for Negative Temperature Coefficient, meaning its resistance decreases as temperature increases. Two to four temperature sensors are placed at each of the positive and negative terminals, as these are the main paths through which current flows and where temperature changes are most drastic, hence the higher weighting coefficients. In the middle of the battery pack, one temperature sensor is placed every 3 to 5 individual cells along the cell arrangement direction, for a total of 6 to 10 sensors, to monitor the internal temperature distribution of the battery pack. Two to four temperature sensors are placed on the outer surface of the battery pack to monitor the impact of ambient temperature on the battery. This multi-point temperature acquisition constructs a three-dimensional temperature field model, which, compared to traditional single-point or dual-point temperature measurements, more accurately reflects the true temperature state of the battery pack, avoiding improper charging parameter settings due to localized overheating or overcooling.

[0028] In step S14, the travel speed is measured by a travel speed sensor mounted on the travel wheel, the steering angle is measured and the steering angular velocity is calculated by a steering angle displacement sensor mounted on the steering mechanism 111, and the hydraulic system power is calculated by a pressure sensor and a flow sensor mounted on the hydraulic system oil pipes. The travel speed sensor is preferably a Hall effect wheel speed sensor or an optical encoder, mounted on the travel wheel axle of the four-wheel drive chassis 11, and calculates the travel speed by detecting the pulse signal generated by the rotation of the wheel axle. The steering angle displacement sensor is preferably a rotary encoder or an angle potentiometer, mounted on the steering shaft of the steering mechanism 111, and measures the steering angle in real time; the steering angular velocity is obtained by taking the time derivative of the steering angle. Calculating the power of the hydraulic system requires simultaneously measuring the pressure and flow rate of the hydraulic oil. Pressure sensors are installed on the oil pipes of hydraulic mechanism I23 and hydraulic mechanism II25 to measure the working pressure. Flow sensor measures hydraulic oil flow. The power of the hydraulic system is calculated using the formula. Calculated; where The power of the hydraulic system is expressed in kW. Hydraulic pressure, unit: MPa; Hydraulic flow rate, in L / min; For hydraulic system efficiency, a value typically ranges from 0.85 to 0.95. Real-time acquisition of operating parameters provides key features for operating condition identification, enabling charging control to dynamically adjust according to the actual operating status of the platform.

[0029] In step S15, a six-axis attitude sensor installed at the center of the walking platform 1 measures the platform's tilt angle and vibration acceleration in the X, Y, and Z directions. The six-axis attitude sensor integrates a three-axis accelerometer and a three-axis gyroscope. The accelerometer measures acceleration in three orthogonal directions, and the gyroscope measures angular velocity along the three rotation axes. By fusing the data from the accelerometer and gyroscope, and using a complementary filtering algorithm or a Kalman filtering algorithm, the platform's pitch, roll, and yaw angles can be accurately calculated. The pitch angle corresponds to the fore-and-aft tilt angle. The roll angle corresponds to the left and right yaw angles. A six-axis attitude sensor is installed at the geometric center of the chassis of the walking platform 1 to ensure that the measurement results represent the attitude of the entire platform; the sensor connects to I... 2 Communicating with the main controller via C bus or SPI bus, I 2 C stands for Inter-Integrated Circuit, a two-wire serial bus; SPI stands for Serial Peripheral Interface, a high-speed synchronous serial communication interface. Vibration acceleration data is used to detect the platform's vibration status during operation. Vibration occurs when the platform travels on uneven ground or when the boom extends or retracts. Excessive vibration may cause short circuits within the battery pack or loose connections, therefore, charging parameters need to be adjusted based on the vibration conditions. Real-time monitoring of attitude and vibration information provides crucial information for safe charging protection.

[0030] In step S16, all collected parameters are transmitted to the main controller via the CAN bus. The main controller performs validity verification and filtering on the data. The CAN bus adopts a multi-master contention bus structure, with each acquisition module acting as a CAN node connected to the bus via twisted-pair cables. Each node sends a data frame containing an identifier, a data length code, and a data field. The receiving node determines whether to receive the data frame based on the identifier. The main controller acts as the master node, periodically polling each sensor node or having the sensor nodes actively report data. Data validity verification includes range checking, rate of change checking, and CRC checking. CRC stands for Cyclic Redundancy Check, used to detect data transmission errors. For data exceeding a reasonable range, the main controller marks it as invalid data and replaces it with the previous valid data or a default value. Filtering uses digital filtering algorithms such as moving average filtering or Kalman filtering to eliminate noise interference in the collected data. The filtering window length is determined based on the signal characteristics. For slowly changing temperature signals, a longer filtering window, such as 10 to 20 sampling points, can be used; for rapidly changing current signals, a shorter filtering window, such as 3 to 5 sampling points, is used. The quality of the data after verification and filtering is guaranteed, providing a reliable input for subsequent control algorithms.

[0031] In step S2, a working condition feature vector is constructed to identify the current working condition type. Working condition identification is one of the core innovations of this invention. By intelligently identifying the working status of the platform, the charging strategy can be adaptively adjusted, avoiding the problem that traditional fixed charging modes cannot adapt to changes in working conditions.

[0032] In step S21, a seven-dimensional working condition feature vector is constructed: .in Walking speed, in km / h; The turning angular velocity is expressed in ° / s. The power of the hydraulic system is expressed in kW. The tilt angle of the platform is expressed in degrees (°). Vibration acceleration, unit is m / s² 2 ; This refers to the current driving the motor, measured in amperes (A). This refers to the boom load power, measured in kW. Travel speed. Directly obtained from the walking speed sensor; steering angular velocity The steering angle is obtained by differentiating the steering angle with respect to time from the steering angle displacement sensor; hydraulic system power. The platform tilt angle has been calculated using pressure and flow sensors in step S14. Pitch angle measured by a six-axis attitude sensor and left and right tilt angles The calculation is obtained, and the calculation formula is: This formula combines the tilt angles in two directions into a comprehensive tilt angle; vibration acceleration Three-axis acceleration from a six-axis attitude sensor , , The calculation is obtained, and the calculation formula is: This formula calculates the vector sum of accelerations in three directions; drive motor current. The data read from the drive controller of electric pump 14 reflects the power requirements of the traveling platform; boom load power. These seven features are calculated based on the positions and speeds of the main telescopic boom 22 and the insulating boom 24, as well as the load on the working bucket 27. They characterize the platform's operating status from different dimensions and constitute the input vector for operating condition identification.

[0033] In step S22, the seven-dimensional working condition feature vector is normalized using the following formula: .in is the normalized value of the i-th feature; This represents the actual measured value of the i-th feature; The minimum value of the i-th feature; This represents the maximum value of the i-th feature. Normalization maps features with different dimensions and numerical ranges to a uniform range of 0 to 1, eliminating the impact of differences in dimensions and numerical ranges between features on pattern recognition. The minimum and maximum values ​​of each feature are determined based on the platform's design parameters and actual operating data. For example, the minimum walking speed is 0 km / h, and the maximum value is set to 3 to 5 km / h based on the platform's maximum walking speed. The normalized feature vector has the same weight in each dimension, facilitating subsequent neural network processing. Normalization is a standard preprocessing step for neural network inputs, which can accelerate network convergence and improve recognition accuracy.

[0034] In step S23, the normalized feature vector is input into a pre-trained fuzzy neural network, which includes 7 input nodes, 15 to 25 hidden layer nodes, and 7 output layer nodes. The fuzzy neural network combines the advantages of fuzzy logic and neural networks, enabling it to handle uncertain information while also possessing learning capabilities. The seven nodes in the input layer each receive one of the seven components of the normalized feature vector. The number of hidden layer nodes needs to be determined based on the sample size and classification complexity; too few nodes will lead to underfitting, while too many nodes will lead to overfitting. An optimal number of hidden layer nodes is 15 to 25. The seven nodes in the output layer correspond to the following conditions: stationary standby (W1), low-speed walking (W2), high-speed walking (W3), steering maneuvering (W4), boom light-load operation (W5), boom heavy-load operation (W6), and mixed operation (W7). The training process of the fuzzy neural network requires collecting at least 500 measured data points from various operating condition switching cycles as training samples. These training samples should cover various operating condition types and transition scenarios. Training employs the backpropagation algorithm (BP), which adjusts network weights by minimizing the output error. After training, the network parameters are stored in the main controller's program memory. Compared with traditional threshold judgment methods, using fuzzy neural networks for working condition identification can better handle complex situations such as fuzzy working condition boundaries and the coexistence of multiple working conditions.

[0035] In step S24, the output values ​​of each node in the output layer are calculated, and the working condition type corresponding to the node with the largest output value is selected as the identification result, which is identified as one of the following: static standby working condition, low-speed walking working condition, high-speed walking working condition, steering maneuvering working condition, boom light-load operation working condition, boom heavy-load operation working condition, or mixed operation working condition. The output values ​​of each node in the output layer are normalized by the Softmax activation function. The Softmax function converts the output into a probability distribution, and the sum of the output values ​​of each node is 1. Selecting the node with the largest output value means selecting the working condition type with the highest probability. When the maximum output value is less than a preset threshold, such as 0.6, it indicates that the current working condition is not typical or is in the transition stage of working condition switching. At this time, it is identified as a mixed operation working condition. After the working condition identification result is output, it is passed to the charging stage selection unit as an important basis for selecting the charging strategy. Accurate working condition identification is a prerequisite for realizing adaptive charging, so that the charging parameters can be dynamically adjusted according to the working condition.

[0036] In step S3, the charging stage is selected based on the operating condition type and SOC value, and the reference charging current is determined. The charging process is divided into five stages, each employing a different charging strategy, which is dynamically adjusted according to the operating condition type to achieve comprehensive optimization of charging speed, charging efficiency, battery temperature rise, and battery life.

[0037] In step S31, the range of the current SOC value is determined. When the SOC is less than 0.20, the pre-charging stage is selected; when the SOC is between 0.20 and 0.60, the fast charging stage is selected; when the SOC is between 0.60 and 0.85, the optimized charging stage is selected; when the SOC is between 0.85 and 0.95, the equalization charging stage is selected; and when the SOC is greater than or equal to 0.95, the float charging maintenance stage is selected. SOC stands for State of Charge, representing the battery's state of charge or remaining capacity. SOC=0 indicates a fully discharged battery, and SOC=1 indicates a fully charged battery. The SOC value is calculated and output in real time by the SOC estimation unit. The five-stage charging strategy is designed for the characteristics of the operating platform. The pre-charging stage uses a small current to wake up deeply discharged batteries, avoiding damage caused by high-current charging. The fast charging stage uses a larger current to quickly replenish the capacity when the battery voltage is low, shortening the charging time. The optimized charging stage gradually reduces the charging current as the battery voltage rises, reducing battery polarization and heat generation. The equalization charging stage balances the voltage differences of individual cells, improving the consistency of the battery pack. The float charging maintenance stage uses constant voltage float charging when the battery is close to full charge to compensate for self-discharge losses. Compared with the traditional two-stage constant current and constant voltage charging, the five-stage strategy can better adapt to changes in battery characteristics and extend battery life.

[0038] In step S32, the reference charging current is determined based on the selected charging stage and the current operating condition type. During the pre-charging phase, settings are configured for stationary standby, low-speed travel, and high-speed travel conditions. The setting is 0.15C to 0.25C, for steering maneuvering and boom operation conditions. The voltage range is 0.03C to 0.07C; during the fast charging phase, the voltage is set under standby conditions. The current is set to 0.45C to 0.55C. During the walking operation, pulse charging is used and the average current is calculated. During the boom operation, the current is set... The reference charging current ranges from 0.25C to 0.35C. Here, C represents the battery's rated capacity in Ah. For example, for a 50Ah battery pack, 1C = 50A, and 0.2C = 10A. The setting of the reference charging current takes into account both charging speed and battery safety. A larger charging current can be used under low-load conditions such as standby, while a smaller charging current is needed under high-load conditions such as steering maneuvers and boom movements to ensure priority power supply. The smaller current during the pre-charging stage is to protect the deeply discharged battery and prevent excessive lithium-ion concentration gradients that could lead to lithium dendrite formation. The larger current during the fast charging stage is to fully utilize the battery's ability to accept high-current charging at low SOC. During walking operations, pulse charging provides power for walking while charging, and it also reduces battery polarization and improves charging acceptance.

[0039] In step S33, during the optimized charging phase, according to the formula... Calculate the reference charging current; during the equalization charging phase, monitor the voltage difference between individual cells. According to the formula Calculate the reference charging current; during the float charging maintenance phase, constant voltage charging is used, with the charging voltage set between 54.4V and 54.8V. The maximum charging current is 0.4C to 0.5C, in A; SOC is the current state of charge. This represents the difference between the highest and lowest voltages of a single battery cell, expressed in volts (V). The charging current during the optimized charging phase decreases linearly with the state of charge (SOC). When SOC = 0.6... When SOC=0.85 Approaching zero, this decreasing strategy aligns with the characteristic of lithium batteries exhibiting decreased charge acceptance at high SOC levels, thus preventing overcharging. During the equalization charging phase, the current is dynamically adjusted based on the individual cell voltage difference; a larger voltage difference results in a larger equalization current, accelerating the equalization process. Equalization charging ends when the voltage difference is less than 50mV. The constant voltage during the float charging maintenance phase is set at 54.6V, corresponding to a float charging voltage of 4.2V per cell in a 13-cell configuration of a 48V lithium battery system. With constant voltage charging, the charging current naturally decays, and charging is considered complete when the current drops below 0.02C. The precise setting of charging parameters at each stage ensures the safety and efficiency of the charging process.

[0040] In step S34, the operating condition adjustment coefficient is determined according to the operating condition type. In standby mode Low-speed travel conditions Up to 0.95, under high-speed driving conditions Up to 0.85, under steering maneuvering conditions Up to 0.80, under light load operating conditions of the boom Up to 0.90, under heavy-load boom operation conditions Up to 0.70, under mixed operating conditions The coefficient is set to 0.75. The working condition adjustment coefficient reflects the degree of limitation on charging current under different working conditions; a smaller coefficient indicates a greater reduction in charging current. In the stationary standby condition, the platform is unloaded and can charge at full power, hence the coefficient is 1.0. In the traveling condition, some power is needed for driving, so the charging power is reduced accordingly. The steering maneuver condition requires high power responsiveness and needs to reserve more power, thus the charging current is reduced significantly. During heavy boom operation, the hydraulic system consumes a lot of power, requiring a substantial reduction in charging current. The mixed operation condition considers multiple loads and uses a moderately low coefficient. The introduction of the working condition adjustment coefficient allows charging control to be coordinated with operational needs, preventing charging from affecting normal operation.

[0041] In step S35, the adjusted charging current is calculated. .in The adjusted charging current is expressed in amperes (A). This charging current takes into account both the charging stage and the operating condition, serving as the benchmark for subsequent temperature compensation and power distribution. Through operating condition adjustments, the charging current can dynamically adapt to changes in the platform's operating status, achieving intelligent coordination between charging and operation.

[0042] In step S4, a three-dimensional temperature field model is established based on multi-point temperature data, the comprehensive temperature compensation coefficient is calculated, and the charging current is adjusted. Temperature is a key factor affecting the charging performance of lithium batteries. Too low a temperature will lead to a decrease in charging acceptance and an increased risk of lithium dendrite precipitation, while too high a temperature will accelerate battery aging and even cause thermal runaway. Therefore, it is necessary to accurately adjust the charging parameters according to the temperature.

[0043] In step S41, real-time temperature data from all temperature sensors are read to obtain a temperature vector. .in The temperature measured by the i-th temperature sensor is expressed in °C. The number of temperature sensors is determined based on the battery pack size; for a 48V lithium battery pack, 16 temperature sensors are preferred. Temperature data is aggregated by a temperature data acquisition board and transmitted to the main controller via the CAN bus. The temperature sampling frequency is relatively low, preferably 1Hz to 10Hz, because temperature changes relatively slowly. The read temperature data needs to be validated to exclude outliers caused by sensor malfunctions. Multi-point temperature acquisition constructs a temperature distribution image of the battery pack, providing a data foundation for accurate temperature compensation.

[0044] In step S42, position weighting coefficients are determined based on the location of each temperature sensor. The weighting coefficients at the positive and negative terminals are 0.15 to 0.20, those at the middle of the battery pack are 0.10 to 0.15, and those on the surface of the battery pack casing are 0.05 to 0.10. These weighting coefficients are set based on the importance of each temperature measurement point to charging control. The positive and negative terminals are the main current channels, and their temperature has the greatest impact on charging performance, therefore they have the highest weight. The middle of the battery pack is the main heat-generating area and requires close monitoring. The surface temperature of the casing is greatly affected by the environment and is mainly used to judge the heat dissipation effect, so its weight is relatively low. The sum of the weighting coefficients of all temperature measurement points is normalized to 1. The weighting coefficients can be optimized and adjusted based on actual operating data. A reasonable weighting allocation allows the overall temperature compensation coefficient to more accurately reflect the true thermal state of the battery pack.

[0045] In step S43, a single-point temperature compensation factor is calculated for the temperature at each temperature measurement point. .when When the temperature is below -10℃ Up to 0.6, when Between -10℃ and 0℃ to ,when Between 0℃ and 20℃ to ,when Between 20℃ and 35℃ Up to 1.05, when When the temperature is between 35℃ and 50℃ to ,when When the temperature is above 50℃ The single-point temperature compensation factor establishes a mapping relationship between temperature and charging current adjustment coefficient, which is determined based on the temperature characteristics and safety requirements of lithium batteries. In extremely low-temperature environments below -10℃, electrolyte viscosity increases, lithium-ion diffusion rate decreases, and charging acceptance capacity significantly declines. Therefore, the compensation factor is set to 0.4 to 0.6, and the charging current is reduced to 40% to 60% of the normal value. In the range of -10℃ to 0℃, battery performance gradually recovers as temperature rises, and the compensation factor increases linearly. In the range of 0℃ to 20℃, it continues to increase linearly. In the suitable temperature range of 20℃ to 35℃, battery performance is optimal, and the compensation factor is close to 1.0, allowing normal charging. In the high-temperature range of 35℃ to 50℃, to prevent further temperature increases, the compensation factor gradually decreases, limiting the charging current. At excessively high temperatures above 50℃, the compensation factor drops to 0.3 to 0.5, significantly reducing the charging current or even stopping charging. The segmented setting of the temperature compensation factor precisely characterizes the charging characteristics of lithium batteries across the entire temperature range.

[0046] In step S44, the comprehensive temperature compensation coefficient is calculated using a weighted average method. .in This is the comprehensive temperature compensation coefficient; The position weighting coefficient for the i-th temperature measurement point; This represents the summation over all temperature measurement points. The comprehensive temperature compensation coefficient integrates the information from each temperature measurement point into a single adjustment parameter, taking into account both the spatial distribution of temperature and highlighting the temperature at key locations through weighting. Compared to simple arithmetic averages or taking the maximum or minimum value, the weighted average method can more scientifically reflect the overall thermal state of the battery pack.

[0047] In step S45, the maximum temperature gradient within the battery pack is calculated. .in The maximum temperature gradient is expressed in °C / dm. The absolute value of the temperature difference between any two temperature measurement points, in °C; The distance between the two temperature measurement points is expressed in decimeters. The temperature gradient reflects the uniformity of temperature distribution within the battery pack. An excessively large temperature gradient indicates localized overheating or poor heat dissipation, which can lead to inconsistent performance of individual cells and accelerate battery aging. When it exceeds 5℃ / dm, Further reduced to to At the same time, the air-cooling system is activated to enhance heat dissipation; the air-cooling system adjusts the heat dissipation capacity by controlling the speed of the cooling fan; the temperature gradient protection can prevent safety hazards caused by local overheating and ensure that the battery pack works in a uniform temperature environment.

[0048] In step S46, the charging current is corrected according to the comprehensive temperature compensation coefficient, and the temperature-compensated charging current is calculated. .in This is the temperature-compensated charging current, expressed in amperes (A). This charging current, based on operating condition adjustments, further considers temperature factors, achieving an adaptive response to temperature changes. The temperature compensation mechanism is a key measure to ensure safe charging of the battery across its entire temperature range.

[0049] In step S5, an extended Kalman filter and a long short-term memory neural network are used to jointly estimate the State of Charge (SOC) and State of Health (SOH) to correct the charging parameters. Accurate estimation of SOC and SOH is a core function of the battery management system. SOH, short for State of Health, represents the battery's health status and reflects the degree of capacity and performance degradation relative to a new battery. This invention combines the advantages of both algorithms, improving estimation accuracy and robustness.

[0050] In step S51, a second-order RC equivalent circuit model is established, and the state vector is defined. Where SOC(k) is the state of charge at time k; and These represent the voltages of the two RC branches, in V; T denotes matrix transpose; k represents the discrete time step. The second-order RC equivalent circuit model represents the battery as an open-circuit voltage source, internal resistance, and two parallel RC branches. The first RC branch simulates the electrochemical polarization process of the battery, with a smaller time constant; the second RC branch simulates the concentration polarization process, with a larger time constant. This model can well describe the dynamic characteristics of the battery, balancing model accuracy and computational complexity. The state vector contains three state variables, with SOC being the core estimation target. and It is an auxiliary state variable.

[0051] In step S52, the extended Kalman filter algorithm is executed to calculate the prior state estimate. .in For prior state estimation; A represents the posterior state estimate from the previous time step; A is the state transition matrix; B is the input matrix. This represents the charging / discharging current at the previous moment, in amperes (A). EKF is an abbreviation for Extended Kalman Filter, a generalization of Kalman filtering to nonlinear systems. The EKF algorithm consists of a prediction step and an update step. The prediction step predicts the current state based on the system model, and the update step corrects the prediction based on measured values. The state transition matrix A and the input matrix B are determined by the battery model parameters, specifically in the form of... , ;in The sampling period is expressed in seconds (s). and is the time constant of the two RC branches, in seconds; C is the rated capacity of the battery, in Ah. and The resistances of the two RC branches are in Ω; the coefficient 3600 is the conversion factor for converting hours to seconds. The prior estimate provides a preliminary prediction of the current state based on the system model.

[0052] In step S53, the Kalman gain is calculated and the posterior state estimate is updated, and the Kalman gain is extracted from the posterior state estimate. As the SOC estimate for the extended Kalman filter algorithm. Kalman gain. The calculation formula is: ;in Let H be the prior estimation error covariance matrix; H be the observation matrix; R be the measurement noise covariance matrix; the superscript T denotes transpose, and the superscript -1 denotes matrix inversion; the formula for calculating the posterior state estimate is: ;in The measured value at time k is the battery terminal voltage; To estimate the predicted measurement value based on the prior state; The residual, or innovation, reflects the deviation between the actual measured value and the predicted value; the Kalman gain determines the degree to which the innovation corrects the state estimate; the posterior estimate integrates model predictions and actual measurements, and has better resistance to accumulated errors compared to the simple ampere-hour integration method; from the posterior state vector The first component can be obtained The EKF algorithm continuously corrects the SOC estimate through recursive calculation, and has the advantages of fast convergence speed and high computational efficiency.

[0053] In step S54, a feature vector containing 30 consecutive sampling points is constructed. The feature vector includes SOC, voltage, current, temperature, internal resistance, voltage change after resting, and cumulative loop count. The feature vector is input into a trained Long Short-Term Memory (LSTM) neural network, which includes a first LSTM layer, a Dropout layer, a second LSTM layer, and a fully connected layer. The output is... and As estimates of SOC and SOH in the neural network. LSTM stands for Long Short-Term Memory, a special type of recurrent neural network (RNN) capable of learning long-term dependencies; 30 consecutive sampling points constitute the time window, with sampling intervals ranging from 1 to 10 seconds, covering the time scale of the battery's dynamic response; each sampling point of the feature vector contains 7 feature parameters, thus the input tensor dimension is 30×7; internal resistance is calculated using the ratio of voltage to current, reflecting the degree of battery aging; voltage change after rest refers to the recovery of the voltage at the end of the battery after charging and discharging stops, reflecting the degree of battery polarization; the cumulative number of cycles is obtained from battery usage records; the first LSTM layer contains 128 units for extracting temporal features; the dropout rate of the Dropout layer is 0.2, randomly discarding 20% ​​of neurons to prevent overfitting; the second LSTM layer contains 64 units for further extracting high-level features; the fully connected layer contains 32 neurons, and the output layer contains 2 neurons for outputting... and The neural network is trained using historical data. The training set should include data under different temperatures, aging conditions, and operating conditions to ensure the network's generalization ability. LSTM neural networks can capture the complex changes in battery characteristics over time and with aging, overcoming the EKF algorithm's strong dependence on the model.

[0054] In step S55, the confidence variance of the extended Kalman filter and the neural network estimation is calculated, and the weighting coefficients are calculated based on the confidence variance. and .in The confidence variance estimated by extended Kalman filtering; The confidence variance is used to estimate the confidence level for the neural network. The confidence variance reflects the uncertainty of the estimation result; a smaller variance indicates a more reliable estimation. The confidence variance of EKF is derived from the posterior estimation error covariance matrix. Extracted from the LSTM; the confidence variance is obtained by calculating the standard deviation of the output through multiple inferences; the weight coefficients are calculated using the inverse variance weighting principle, with methods estimating smaller variances having larger weights, and methods estimating larger variances having smaller weights; the sum of the two weights is 1, i.e. Dynamic weight allocation achieves the complementary advantages of the two algorithms, increasing the weight of EKF when the reliability of the estimation is high, and increasing the weight of LSTM when the reliability of the estimation is high.

[0055] In step S56, the final SOC and SOH values ​​are calculated using Bayesian fusion, and the calculation formula is as follows: and .in This is a correction factor, ranging from -0.8 to -1.2; This represents the relative change in internal resistance. Bayesian fusion is a probabilistic reasoning method that weights two estimation results to obtain the final result. It combines the model-driven estimation of EKF and the data-driven estimation of LSTM, resulting in higher accuracy and robustness; Based on LSTM estimation, an internal resistance correction term is introduced. Internal resistance is an important parameter reflecting the health status of the battery. An increase in internal resistance usually means that the battery is aging. ,in As for the current internal resistance, Internal resistance of the new battery; correction factor A negative value indicates that the internal resistance increases. The fusion algorithm fully utilizes multi-source information, improving the accuracy of SOC and SOH estimations.

[0056] In step S57, according to the final Value correction charging parameters, when When the current is less than 0.90, the maximum charging current will be reduced to 0.35C to 0.45C. When the value is less than 0.85, the maximum charging current is further reduced to 0.30C to 0.40C. When the voltage is less than 0.80, the charging cutoff voltage will be reduced by 0.03V to 0.08V. Charging parameters will vary accordingly. Adaptive adjustment is a key innovation of this invention. Traditional charging methods use the same parameters for all batteries without considering the impact of battery aging. As batteries age, their internal resistance increases, active materials are lost, and their charging acceptance decreases. If the charging parameters of new batteries are still used, it may lead to overcharging, overheating, or even safety accidents. Reducing the maximum charging current can alleviate the charging burden on aging batteries. Lowering the charging cutoff voltage avoids overcharging, which can accelerate aging. The adaptive strategy extends the battery's full lifespan and improves economic efficiency.

[0057] In step S6, future power demand is predicted, and the power source between the diesel engine 12 and the electric pump 14 is switched to dynamically allocate charging power. The work platform is equipped with dual power sources, a diesel engine and an electric pump. Reasonable power source switching and power allocation can improve energy utilization efficiency and reduce diesel consumption and carbon emissions.

[0058] In step S61, historical power data from the past 5 to 10 minutes is collected to establish a power series, and a sliding window time series analysis method is used to predict the power demand for the next 30 to 60 seconds. The prediction formula is .in For predicted power, the unit is kW; The predicted duration is in seconds (s). Historical data weighting coefficients; Historical power values, in kW; This represents the power trend term, measured in kW. The power series records the platform's power consumption over a period of time, including walking power, hydraulic power, and auxiliary equipment power. Sliding window analysis is a common method for time series forecasting, predicting future values ​​by weighted summation of historical data. The weight coefficients of historical data decrease over time, with recent data having a higher weight and older data having a lower weight; for example, exponentially decaying weights can be used. ,in Attenuation coefficient; power change trend term Obtained by taking the first or second difference of the power sequence, it reflects the direction and speed of power change; the prediction time is set to 30 to 60 seconds, which can make power source switching decisions in advance without increasing the error due to excessive prediction time; short-term load forecasting provides a basis for predictive energy management and avoids the problems of frequent power source switching and insufficient power.

[0059] In step S62, the power source switching conditions are determined. When the load power is greater than 0.80 to 0.90 times the maximum power of the electric pump 14, or the SOC is less than 0.25 to 0.35, or the predicted power is greater than 0.85 to 0.95 times the rated power of the electric pump 14, the power source is switched from the electric pump 14 to the diesel engine 12. When the load power is less than 0.55 to 0.65 times the output power of the diesel engine 12, and the SOC is less than 0.85 to 0.95, and the duration is greater than 8s to 12s, the power source is switched from the diesel engine 12 to the charging mode. Power source switching follows the principles of load priority, efficiency priority, and battery protection. When the load power approaches the electric pump's maximum power, it indicates that the electric pump is about to overload, requiring a switch to the diesel engine to ensure power supply. When the State of Charge (SOC) is too low, it indicates insufficient battery power, necessitating a switch to the diesel engine to drive and charge the battery. When the predicted power exceeds the electric pump's rated power, it indicates that the future load will increase, requiring an advance switch of power source. Switching from diesel engine to charging mode requires three conditions: first, the load power is relatively low, and the diesel engine has surplus power available for charging; second, the SOC is not full, and the battery has a charging need; and third, the low-load state persists for a certain period to avoid frequent switching during brief load drops. The duration judgment uses a time-delay filter, and switching is only executed if the conditions are consistently met for 8 to 12 seconds. This reasonable switching strategy reduces the number of diesel engine start-stop cycles, extends engine life, and ensures the power requirements of the work platform.

[0060] In step S63, in charging mode, the remaining power available for charging is calculated. .in The remaining power available for charging, in kW; The output power of the diesel engine is 12 kW; Load power, in kW; The reserved power is determined based on the predicted power, and the unit is kW. The output power of the diesel engine is read through the engine controller; the load power includes the sum of power consumption of various components such as travel, hydraulics, and control systems; the reserved power is set according to the future power demand predicted in step S61. A certain amount of power is reserved to cope with sudden load increases and avoid insufficient power; the remaining power is used for battery charging, realizing full utilization of energy.

[0061] In step S64, the charging power is calculated. .in Actual charging power, in kW; This indicates taking the minimum value; Maximum charging power, in kW; For charging efficiency, the value ranges from 0.88 to 0.95; This is the power adjustment factor based on SOC; This is a temperature-based power adjustment factor. The charging power cannot exceed the battery's maximum charging power; therefore, a smaller value must be chosen between the remaining power and the maximum charging power. Charging efficiency takes into account energy losses during the charging process, including conversion losses from the charging module and charging losses from the battery. As the State of Charge (SOC) changes, a higher charging power can be accepted when the SOC is lower, while a lower charging power needs to be accepted when the SOC is higher. For example... ; Adjust the charging power according to the battery temperature; reduce the charging power when the temperature deviates from the optimal range, for example... The precise calculation and dynamic adjustment of charging power enable optimal control of the charging process.

[0062] In step S65, the charging current is calculated based on the charging power and battery voltage. .in The charging current is calculated based on power allocation, and the unit is ampere-ampere (A). This is the current battery pack voltage, in volts (V). The battery voltage is measured in real time by a voltage acquisition module; the charging current is calculated based on the relationship between power and voltage; this charging current reflects the energy management requirements, and is taken as the smaller value compared to the charging current calculated earlier based on the charging stage and operating conditions.

[0063] In step S66, when the boom 22 descends or the vehicle decelerates, it is detected whether the drive motor is in a generating state. If it is in a generating state and the energy recovery conditions are met, the energy recovery mode is activated, the energy recovery current is calculated, and it is limited to the range of 0.3C to 0.6C. Energy recovery is an important means to improve energy utilization efficiency. When the boom descends, gravitational potential energy is converted into electrical energy, and when the vehicle decelerates, kinetic energy is converted into electrical energy. Whether the drive motor is in a generating state can be determined by the working mode register of the motor controller or by detecting the direction of the current. The energy recovery conditions include SOC less than 0.95%, battery temperature less than 45°C, and recovery current not exceeding 0.5C. The energy recovery current is calculated based on the boom descent speed or vehicle deceleration, and the calculation formula is as follows: .in This is the energy recovery current, measured in amperes (A). Energy recovery efficiency, ranging from 0.70 to 0.85; m is the mass of the boom or load, in kg; g is the acceleration due to gravity, taken as 9.8 m / s². 2 h represents the boom descent height, in meters (m). The descent time is expressed in seconds (s); J represents the moment of inertia, expressed in kg·m. 2 ; The value is angular velocity, measured in rad / s. The recovery current is limited to the range of 0.3C to 0.6C to protect the battery and prevent damage caused by excessive charging current. The application of energy recovery mode reduces overall energy consumption and extends the working time after a single charge.

[0064] In step S7, the platform tilt angle and vibration acceleration are detected, and charging is adjusted or stopped. Aerial work platforms tilt and vibrate on uneven ground or when the boom is extended, which affects battery charging safety and requires appropriate protective measures.

[0065] In step S71, the platform's forward and backward tilt angles are read from the six-axis attitude sensor. and left and right tilt angles Calculate the overall tilt .in The total tilt is expressed in degrees (°). The angle is the forward or backward tilt angle, with forward tilt being positive and backward tilt being negative, and the unit is °; The tilt angles are left and right, with rightward tilt being positive and leftward tilt being negative, in degrees. The overall tilt angle is calculated by vector synthesis to combine the effects of the two tilt angles; the tilt angle data is read in real time from a six-axis attitude sensor, with an update frequency of 10Hz to 100Hz; platform tilt can cause uneven distribution of electrolyte within the battery pack, affecting the charging status of individual cells, and severe tilt may pose a safety hazard.

[0066] In step S72, the charging current is adjusted according to the overall tilt angle. The charging current remains constant when the temperature is less than 3°. The charging current is adjusted when the temperature is between 3° and 5°. to ,when The charging current is adjusted when the temperature is between 5° and 8°. to ,when Charging will stop and an audible and visual alarm will be triggered when the temperature reaches or exceeds 8°C. The charging current after tilt adjustment is expressed in amperes (A). A tilt angle less than 3° is within the normal range; this range is applicable to most ground surfaces and working postures, and charging is unaffected. A tilt angle between 3° and 5° is considered a slight tilt, requiring a reduction in the charging current, preferably to 85% to 95% of the original value. A tilt angle between 5° and 8° is considered a moderate tilt, requiring a significant reduction in the charging current to 65% to 75% of the original value. A tilt angle greater than or equal to 8° is considered a severe tilt, indicating an abnormal stress state on the battery pack. Charging must be stopped immediately, and an audible and visual alarm should be triggered to prompt the operator to adjust the platform's posture. The alarm signal is output through the buzzer and indicator light on control box 31. After charging stops, the system enters a fault protection state, recording the abnormal event. Charging can only resume after the tilt angle returns to normal and the operator confirms. This tilt protection strategy effectively prevents potential charging safety hazards under tilted conditions.

[0067] In step S73, the three-axis acceleration data of the six-axis attitude sensor are read. , and Calculate the overall vibration acceleration .in The total vibration acceleration is expressed in m / s². 2 ; , , These are the acceleration components in the X, Y, and Z directions, respectively, with units of m / s². 2 The overall vibration acceleration is calculated by the vector sum of accelerations in three directions, reflecting the overall vibration intensity of the platform. Acceleration data is read from the accelerometer of the six-axis attitude sensor. The platform vibration mainly originates from walking on uneven ground, rapid boom movements, engine vibration, etc. Vibration can cause problems such as internal structural stress of the battery pack, loose connections, and electrolyte sloshing.

[0068] In step S74, the charging current is adjusted according to the comprehensive vibration acceleration. If the amount exceeds 0.4g to 0.6g and lasts for more than 2s to 4s, reduce the charging current to [a lower value]. to ,when Charging will be paused immediately if the weight exceeds 0.8g but is between 1.2g. Charging will resume 3-7 seconds after the vibration returns to normal. The current is the charging current after vibration adjustment, in amperes (A); g is the unit of gravitational acceleration, 1g = 9.8 m / s². 2 Mild vibration acceleration between 0.4g and 0.6g requires a certain duration to trigger protection, preventing malfunctions caused by instantaneous vibration. During moderate vibration, the charging current drops significantly to 25% to 35% of its original value, reducing heat and stress during charging. Severe vibration acceleration exceeding 0.8g to 1.2g indicates the platform is under severe vibration, necessitating immediate charging halt. Even after charging is paused and vibration returns to normal, charging resumes only after a 3-7 second delay. This delay ensures the vibration has truly stopped, rather than experiencing brief fluctuations; the delayed resumption also provides time for the battery pack to stabilize. This vibration protection strategy reduces the risk of internal short circuits and connection failures caused by vibration.

[0069] In step S75, the voltage fluctuation of the charging circuit is monitored, and the voltage fluctuation rate is calculated. .in Voltage fluctuation rate; The maximum voltage within the sampling window, in V; This is the minimum voltage, measured in volts (V). The average voltage is expressed in volts (V). The sampling window is set to 1 to 5 seconds, containing multiple voltage sampling points. Voltage fluctuation reflects the stability of the charging circuit voltage; voltage fluctuation is minimal during normal charging. If the current exceeds 0.04 to 0.06, the connector is considered potentially loose. The charging current is reduced to 0.10C to 0.20C, and an alarm is triggered. A loose connector leads to increased contact resistance and unstable contact, manifesting as increased voltage fluctuations. Causes of loose connectors include vibration, thermal expansion and contraction, and improper installation. Reducing the charging current can decrease heat generation and arcing in loose connectors. The alarm signal alerts operators to check the connector and address the looseness issue promptly. Connector monitoring is a crucial aspect of electrical safety, preventing overheating and fire risks caused by poor contact.

[0070] In step S76, the smaller value between the tilt-adjusted charging current and the vibration-adjusted charging current is selected as the final safe charging current and sent to the charging controller for execution. Final safe charging current ;in This is the final charging current, expressed in amperes (A). This represents the minimum value function. Taking the smaller value follows the safety priority principle, ensuring that the charging current simultaneously meets the safety requirements of both tilt and vibration. The final charging current integrates multiple factors such as charging stage, operating condition adjustment, temperature compensation, power distribution, tilt protection, and vibration protection, and is the result of coordinated optimization of each link. This charging current is sent to the charging controller of the charging execution unit, and the charging controller controls the output current of the charging module to execute this instruction.

[0071] In step S8, charging and discharging data is uploaded to the cloud server, the optimized charging strategy parameters are received, and the local algorithm is updated. Cloud collaboration is an important means to achieve continuous optimization and predictive maintenance of charging strategies, and the charging control algorithm is continuously improved through big data analysis and machine learning.

[0072] In step S81, after each operation, charge / discharge curve data, historical temperature data, estimated SOC and SOH values, operating condition type statistics, abnormal event records, and environmental parameters are read from the data storage module. The data storage module uses a FLASH memory or SD card, with capacity determined based on data volume requirements. The charge / discharge curve data includes voltage and current change curves over time, recorded at intervals of 1 to 10 seconds. Historical temperature data includes temperature changes at each measurement point over time. The operating condition type statistics record the duration and number of switching cycles for each operating condition. Abnormal event records include the occurrence time and parameters of events such as tilt alarms, vibration alarms, temperature anomalies, and voltage anomalies. Environmental parameters include ambient temperature, humidity, and altitude information, which can be obtained from meteorological sensors or a GPS module. This comprehensive data collection provides rich material for cloud-based analysis.

[0073] In step S82, the data is packaged into a data packet in JSON or XML format. The data packet contains fields for vehicle identification number (VIN), operation date, battery data, operating condition data, abnormal events, and environmental data. JSON is short for JavaScriptObjectNotation, a lightweight data exchange format; XML is short for eXtensible Markup Language. Both formats have good readability and extensibility. The VIN is short for VehicleIdentificationNumber, which uniquely identifies the operating platform. The operation date records the time the data was generated. Each field is organized in key-value pairs; for example, battery_data contains subfields such as SOC, SOH, voltage, and current. The data packaging process includes formatting, compression, and encryption. Compression reduces the amount of data transmitted, and encryption protects data security.

[0074] In step S83, data packets are uploaded to the cloud server using the HTTPS protocol via a 4G / 5G wireless communication module or a WiFi module. The 4G / 5G communication module provides wide area network connectivity, suitable for outdoor operation scenarios; the WiFi module provides local area network connectivity, suitable for scenarios where the parking location has WiFi coverage. HTTPS stands for Hypertext Transfer Protocol Secure, based on the HTTP protocol and encrypted using SSL / TLS. SSL stands for Secure Sockets Layer, and TLS stands for Transport Layer Security; both are encrypted communication protocols. The HTTPS protocol ensures the confidentiality, integrity, and authentication of data transmission. The cloud server address is configured via a domain name, supporting domain name resolution and IP address access. Data upload is performed as a background task, not affecting the real-time performance of charging control. If an upload fails, the data is saved locally, awaiting the next upload.

[0075] In step S84, the cloud server receives data from multiple operating platforms and uses a big data processing framework to perform similar operating condition clustering analysis, charging strategy effectiveness evaluation, battery life prediction model training, and optimal charging parameter statistical analysis. The cloud server possesses powerful computing and storage capabilities, enabling it to handle massive amounts of data; big data processing frameworks such as Hadoop and Spark provide distributed computing capabilities; similar operating condition clustering analysis categorizes data from different vehicles under similar operating conditions, facilitating comparative analysis; the charging strategy effectiveness evaluation calculates indicators such as charging time, temperature rise, efficiency, and lifespan under different charging parameters; the battery life prediction model employs machine learning algorithms such as Support Vector Machine (SVM), Random Forest (RF), or deep learning models, taking historical battery data and current state as input and outputting the remaining usable life (RUL); RUL is an abbreviation for Remaining Useful Life; and the optimal charging parameter statistical analysis identifies the optimal parameter combinations under various conditions through statistical analysis of large amounts of data. The cloud analysis uncovers patterns and knowledge within the data, guiding the optimization of charging strategies.

[0076] In step S85, the cloud server generates optimized charging strategy parameters based on big data analysis results. These parameters include current setpoints for each charging stage, charging cutoff voltage, equalization charging time, temperature compensation coefficient matrix, and operating condition adjustment coefficients. The optimized parameters improve charging performance and battery life compared to the initial parameters. Parameter optimization employs a multi-objective optimization algorithm, with optimization objectives including minimizing charging time, minimizing temperature rise, maximizing charging efficiency, and maximizing battery life. Optimization constraints include charging current limits, temperature limits, and SOC range. The optimization algorithm can employ Genetic Algorithm (GA), Particle Swarm Optimization (PSO), or gradient descent. The optimization results are verified through simulation and reviewed by experts before being pushed to the vehicle. The temperature compensation coefficient matrix provides compensation coefficients for different temperature points, and the operating condition adjustment coefficients provide adjustment coefficients for different operating condition types.

[0077] In step S86, the cloud server packages the optimized charging strategy parameters into a configuration file and pushes it to the job platform. The configuration file is in JSON or XML format and includes information such as version number, effective time, and parameter content; the version number is used for parameter management and rollback; the effective time allows parameters to take effect at a specified time, facilitating unified updates; the push adopts a publish-subscribe pattern, with the cloud server as the publisher and the job platform as the subscriber; push messages are sent via the MQTT or WebSocket protocol; MQTT is an abbreviation for MessageQueuingTelemetryTransport, a lightweight message transmission protocol; the job platform's wireless communication module receives the push messages and stores the configuration file.

[0078] In step S87, the main controller of the operating platform receives the configuration file, performs integrity verification and version compatibility checks, and then updates the parameters of the charging stage selection unit, temperature compensation processing unit, and energy management unit. Integrity verification compares the hash value or checksum of the configuration file with the checksum sent from the cloud to ensure the file is not corrupted. Version compatibility checks ensure that the new parameters are compatible with the current software version, avoiding errors caused by parameter mismatches. Parameter updates use atomic operations to ensure parameter consistency. The update process is logged for easy traceability. Functional testing is performed after the parameter update to confirm that the charging control is working properly. The parameter update enables remote upgrades of the charging strategy, i.e., OTA updates. OTA stands for Over-The-Air, referring to software or parameter updates performed wirelessly.

[0079] In step S88, the cloud server predicts the remaining battery life (RUL) and failure probability based on battery aging trends. When the RUL is less than a preset threshold or the failure probability is greater than a preset threshold, an early warning is generated and pushed to the operation platform and management platform. Battery aging trends are obtained through time-series analysis of parameters such as SOH, capacity, and internal resistance. The RUL prediction model takes into account current SOH, charge / discharge cycle count, usage time, and historical temperature characteristics, and outputs the estimated remaining days of use or cycle count. The failure probability prediction model is trained based on historical failure data to identify early warning features. The preset threshold is determined according to the maintenance strategy; for example, an early warning is triggered when the RUL is less than 30 days or the failure probability is greater than 20%. The early warning information includes the warning type, severity, cause analysis, and maintenance recommendations. Early warnings pushed to the operation platform are displayed on the control box screen and trigger audio-visual alerts. Early warnings pushed to the management platform are viewed by maintenance personnel who then schedule maintenance plans. Predictive maintenance avoids downtime losses caused by sudden failures and reduces maintenance costs. Cloud collaboration enables intelligent charging strategies and improves the maintainability of the charging system.

[0080] Through the detailed implementation of steps S1 to S8 above, the battery pack charging control method for the narrow-body self-propelled work platform of the present invention achieves intelligent control of the entire process, including parameter acquisition, working condition identification, stage selection, temperature compensation, SOC / SOH estimation, power allocation, safety protection, and cloud optimization. Compared with traditional charging methods, this method has significant advantages: charging time is shortened by 15% to 25%, charging efficiency is increased by 6% to 9%, battery temperature rise is reduced by 8°C to 12°C, battery cycle life is extended by 20% to 30%, it has strong adaptability to cope with various working conditions and the entire temperature range, high safety with multiple protection mechanisms against tilt, vibration, and temperature, high intelligence with autonomous learning and continuous optimization, and good economic benefits, reducing operating costs by 15% to 20%. This invention provides a complete battery charging management solution for narrow-body self-propelled work platforms, meeting the high requirements for equipment reliability and operational efficiency in power distribution line operations in narrow and complex environments.

[0081] Example 1 describes a lithium battery pack charging control system for a narrow-body self-propelled work platform used to implement the method of the present invention, including a walking platform 1, a work device 2, and a control system 3. The walking platform 1 has a wheeled four-wheel drive chassis 11, a diesel engine 12 mounted in the center of the chassis, a diesel tank 13 located at the rear of the wheeled four-wheel drive chassis 11 and connected to the diesel engine 12 via an oil pipe, an electric pump 14 located at the front of the wheeled four-wheel drive chassis 11, and a battery pack 15 located below or on the side of the wheeled four-wheel drive chassis 11. The working device 2 is mounted on the traveling platform 1 and includes a turntable 21, a main telescopic boom 22, a hydraulic mechanism I 23, an insulated arm 24, a hydraulic mechanism II 25, a slewing mechanism 26, and a work bucket 27. The turntable 21 is mounted on a wheeled four-wheel drive chassis 11 via a slewing bearing. The main telescopic boom 22 is hinged to the turntable 21 and achieves undulating and telescopic movements through the hydraulic mechanism I 23. The insulated arm 24 is hinged to the end of the main telescopic boom 22 and achieves undulating movements through the hydraulic mechanism II 25. The work bucket 27 is connected to the end of the insulated arm 24 via the slewing mechanism 26. The control system 3 includes a control box 31 and a remote controller 32. The control box 31 is mounted on the support of the turntable 21, and the remote controller 32 communicates with the wireless receiving module inside the control box 31 via wireless signals.

[0082] The control box 31 houses a main controller, a parameter acquisition unit, a working condition identification unit, a charging stage selection unit, a temperature compensation processing unit, a SOC / SOH joint estimation unit, an energy management unit, a safety protection unit, a charging execution unit, and a cloud collaboration unit. The parameter acquisition unit acquires the voltage of each individual cell in the battery pack 15 through a voltage acquisition module, the charging and discharging current through a current sensor, the temperature at multiple points within the battery pack 15 through a temperature sensor array, the walking speed through a walking speed sensor, the steering angle through a steering angle displacement sensor and calculates the steering angular velocity, the hydraulic system power through hydraulic pressure and flow sensors, and the platform tilt angle and vibration acceleration through a six-axis attitude sensor. All parameters are transmitted to the main controller via a CAN bus. The working condition identification unit constructs a seven-dimensional working condition feature vector and inputs it into a fuzzy neural network to identify the current working condition as one of the following: static standby, low-speed walking, high-speed walking, steering maneuvering, light-load boom operation, heavy-load boom operation, or mixed operation. The working condition identification result is transmitted to the charging stage selection unit via a CAN bus.

[0083] The charging stage selection unit selects the charging stage based on the operating condition and SOC value, including pre-charging, fast charging, optimized charging, equalization charging, and float charging maintenance stages. It determines the reference charging current, calculates the adjusted charging current using an operating condition adjustment coefficient, and transmits the adjusted charging current to the temperature compensation processing unit via the data bus. The temperature compensation processing unit establishes a three-dimensional temperature field model based on multi-point temperature data, calculates a comprehensive temperature compensation coefficient, and corrects the charging current accordingly. The temperature-compensated charging current is transmitted to the charging execution unit via a signal line. The SOC / SOH joint estimation unit uses extended Kalman filtering and a long short-time memory neural network to jointly estimate SOC and SOH, and calculates the final SOC and SOH values ​​through Bayesian fusion. The final SOC and SOH values ​​are transmitted to the charging stage selection unit, energy management unit, and cloud collaboration unit via the CAN bus.

[0084] The energy management unit collects historical power data and predicts future power demand. Based on load power, SOC value, and predicted power, it switches the power source between the diesel engine 12 and the electric pump 14. In charging mode, it calculates the remaining power available for charging and dynamically allocates charging power. When the boom lowers or the vehicle decelerates, it activates the energy recovery mode. Charging power commands are transmitted to the charging execution unit via the CAN bus. The safety protection unit reads the platform tilt angle and vibration acceleration from the six-axis attitude sensor. Based on the combined tilt angle and vibration acceleration, it adjusts the charging current or stops charging. It monitors voltage fluctuations in the charging circuit and reduces the charging current when connectors become loose. Finally, the safe charging current is transmitted to the charging execution unit via a signal line. The charging execution unit includes a charging controller and a charging module. The charging controller integrates the charging current commands output from each unit to control the output current of the charging module. The input terminal of the charging module is selectively connected to the generator output terminal of the diesel engine 12 or the power output terminal of the electric pump 14 via a power switching switch. The output terminal of the charging module is connected to the battery pack 15 via a charging cable.

[0085] After each operation, the cloud-based collaborative unit reads charging and discharging data from each unit, packages it into data packets, and uploads them to the cloud server via a wireless communication module. It also receives optimized charging strategy parameters pushed by the cloud server and updates the parameters of each unit accordingly. The main controller communicates bidirectionally with each unit via a CAN bus, coordinating data transmission and control command delivery. The main controller's power input is connected to the auxiliary power output of battery pack 15 via a DC-DC converter. The control box 31's panel features a display screen, buttons, indicator lights, and a buzzer to display status information and provide an operating interface.

[0086] Example 2: This example uses the narrow-body self-propelled work platform from Example 1 to perform routine inspection and maintenance of a 10kV power distribution line in a city alley. The work period is from 8:00 AM to 5:00 PM, with an ambient temperature of 25 degrees Celsius. The work platform is equipped with a 48V 50Ah lithium iron phosphate battery pack. Before the work begins, the initial SOC of the battery pack is 45%. The main controller reads parameters such as battery voltage, current, and temperature through the parameter acquisition unit. The SOC-SOH joint estimation unit estimates the current SOH to be 96%. The work platform travels approximately 800 meters from the parking point to the work site. Based on characteristic parameters such as walking speed of 1.2 km / h, turning angular velocity of 0 degrees / s, hydraulic system power of 0.2 kW, platform tilt angle of 1 degree, vibration acceleration of 0.15g, drive motor current of 12 amps, and boom load power of 0 kW, the working condition identification unit identifies the current working condition as a low-speed walking condition using a fuzzy neural network processor. The charging stage selection unit determines that the current charging stage is in progress based on a SOC value of 45%. It then determines the reference charging current as 0.5C, or 25 amps, using a table. Based on low-speed driving conditions, it sets the adjustment factor at 0.90, resulting in an adjusted charging current of 22.5 amps. The temperature compensation processing unit reads data from 16 temperature sensors: positive terminal temperature 26 degrees Celsius, negative terminal temperature 26 degrees Celsius, average temperature in the middle of the battery pack 25 degrees Celsius, and outer casing surface temperature 24 degrees Celsius. The single-point temperature compensation factor for each measurement point is 1.0, and the overall temperature compensation coefficient is calculated to be 1.0. The temperature-compensated charging current remains at 22.5 amps. The energy management unit detected that the diesel engine 12 was idling, with an output power of 3 kW, a current load power of 1.2 kW, a reserved power of 0.3 kW, and a remaining power of 1.5 kW available for charging. The charging power distribution module calculated a charging power of 1.35 kW, a charging efficiency of 0.90, and a charging current of 25.3 amps based on a battery voltage of 48 volts. Comparing this with the previously calculated 22.5 amps, the smaller value was selected, and the final charging current was determined to be 22.5 amps. The safety protection unit detected that the platform tilt angle was less than 3 degrees (1 degree), the comprehensive vibration acceleration was less than 0.4g (0.15g), and the charging circuit voltage fluctuation rate was less than 0.04 (0.02). The safety protection unit output a safe charging current of 22.5 amps. The charging controller, integrating the outputs of all units, controlled the charging module to charge the battery pack 15 with a current of 22.5 amps.

[0087] After a 10-minute drive, the system arrives at the work site and switches to a standby mode, with the charging current adjusted to 25 amps. The operator uses remote control 32 to raise the main telescopic boom 22 to a height of 12 meters. The work bucket 27 reaches the guide wire position, and the work condition identification unit recognizes it as a heavy-load boom operation. The work condition adjustment coefficient is reduced to 0.65, and the charging current is reduced to 16.3 amps to ensure priority power supply for the operation. The operator performs guide wire inspection and tightening within the work bucket 27 for 20 minutes, during which the boom operates under a light load, and the charging current remains around 20 amps. After the operation is completed, the main telescopic boom 22 descends. The energy management unit detects that the drive motor is generating power and activates the energy recovery mode, recovering 12 amps of current for 30 seconds, recovering approximately 0.16 amp-hours of energy. The operation platform continuously completed operations at 8 inspection points within the alley, with a total operation time of 6 hours and a total charging time of 4.5 hours. The SOC was charged from 45% to 88%, with a charging capacity of 21.5 amp-hours. The highest battery pack temperature reached 38 degrees Celsius, with an average temperature rise of 13 degrees Celsius and a charging efficiency of 91%. After the operation, the cloud-based collaborative unit uploaded the charging and discharging data to the cloud server via a 4G wireless communication module. The uploaded data packet was 1.2 megabytes in size and included information such as charging and discharging curves, temperature history, operating condition statistics, and abnormal events. The cloud server analyzed the data and generated optimized charging strategy parameters, which were then pushed to the operation platform. The main controller received the configuration file and updated the current setting value of the charging stage selection unit, adjusting the maximum current of the fast charging stage from 0.5C to 0.52C and the initial SOC of the optimized charging stage from 0.60 to 0.58. It is estimated that the charging time for the next operation can be shortened by 8 minutes.

[0088] To verify the beneficial effects of the charging control method of this invention, comparative examples and verification schemes for experiments 1-5 were designed. Comparative example 1 uses a traditional constant current and constant voltage charging method. During the constant current phase, the charging current is 0.5C, i.e., 25 amps. After charging to a voltage of 54.6 volts, it switches to constant voltage charging, without considering changes in operating conditions or temperature compensation. Comparative example 2 uses a traditional single-point temperature compensation charging method, placing only one temperature sensor at the center of the battery pack and adjusting the charging current based on this temperature, without considering operating condition identification or power prediction. Example 2 employs the complete charging control method of this invention, including operating condition identification, multi-stage adaptive charging, multi-point temperature compensation, joint estimation of SOC and SOH, power prediction and allocation, and safety protection.

[0089] The experimental verification scheme was designed in accordance with GB / T31467.3-2015 "Lithium-ion Power Battery Packs and Systems for Electric Vehicles - Part 3: Safety Requirements and Test Methods" and GB / T31484-2015 "Cycle Life Requirements and Test Methods for Power Batteries for Electric Vehicles". The experimental subjects were three identical 48V 50Ah lithium iron phosphate battery packs, numbered A (using Comparative Example 1), B (using Comparative Example 2), and C (using Example 2). The experiments were conducted in a temperature-controlled laboratory, with ambient temperatures controlled at four points: -10°C, 10°C, 30°C, and 50°C. Charge and discharge performance tests were performed at each temperature point. Charging performance tests were conducted according to the standard charging procedure specified in GB / T31467.3, charging the battery packs from 20% SOC to 95% SOC, and recording the charging time, maximum temperature rise, charging efficiency, and SOC estimation error. Charging time was recorded using a timer; the maximum temperature rise was obtained by measuring the battery surface temperature using a thermocouple temperature sensor; charging efficiency was calculated by dividing the charging output energy by the charging input energy; and the SOC estimation error was obtained by comparing the baseline SOC calculated using the ampere-hour integration method with the SOC estimated by each method. The specific experimental procedure is as follows: In Experiment 1, three identical 48V 50Ah lithium iron phosphate battery packs were numbered A, B, and C. Group A used the traditional constant current and constant voltage charging method of Comparative Example 1, Group B used the single-point temperature compensation charging method of Comparative Example 2, and Group C used the complete charging control method of the present invention as described in Example 2. The experiment was conducted in a temperature-controlled laboratory with ambient temperatures set at four points: -10°C, 10°C, 30°C, and 50°C. At each temperature point, the battery pack was charged from 20% SOC to 95% SOC. Charging time, maximum temperature rise on the battery surface, charging efficiency, and SOC estimation error were recorded. Charging efficiency was calculated by dividing the charging output energy by the charging input energy. The SOC estimation error was obtained by comparing the baseline SOC value calculated by the ampere-hour integration method with the SOC values ​​estimated by each method. Figure 1 The four subplots in the group show the complete results of the charging performance comparison test at different temperatures in Experiment Example 1.

[0090] from Figure 1The charging time comparison curves show that at a low temperature of -10 degrees Celsius, Comparative Example 1 requires 85 minutes to complete charging, Comparative Example 2 requires 78 minutes, while Example 2 only requires 72 minutes. Example 2 shortens the charging time by 15.3% compared to Comparative Example 1. As the ambient temperature rises to 10 degrees Celsius, the charging time of Comparative Example 1 decreases to 65 minutes, Comparative Example 2 to 60 minutes, and Example 2 further to 55 minutes. Under a suitable temperature of 30 degrees Celsius, Comparative Example 1 requires 55 minutes, Comparative Example 2 requires 52 minutes, and Example 2 only requires 46 minutes, maintaining its charging speed advantage. Under a high temperature of 50 degrees Celsius, Comparative Example 1 takes 48 minutes, Comparative Example 2 takes 45 minutes, and Example 2 takes 42 minutes. The data leads to the conclusion that the charging control method of Example 2 achieves the shortest charging time across the entire temperature range, with a particularly significant advantage in low-temperature environments. This is because Example 2 employs a multi-point temperature compensation and adaptive multi-stage charging strategy, precisely adjusting charging parameters according to temperature.

[0091] from Figure 1 The highest temperature rise comparison curves show that temperature control is a key indicator of charging safety. At -10 degrees Celsius, the highest temperature rise of Comparative Example 1 is 8 degrees Celsius, Comparative Example 2 is 7 degrees Celsius, and Example 2 is 6 degrees Celsius. At 10 degrees Celsius, the temperature rise of Comparative Example 1 reaches 12 degrees Celsius, Comparative Example 2 is 10 degrees Celsius, and Example 2 is controlled at 9 degrees Celsius. At 30 degrees Celsius, the temperature rise of Comparative Example 1 increases to 15 degrees Celsius, Comparative Example 2 is 13 degrees Celsius, and Example 2 is 11 degrees Celsius. At 50 degrees Celsius, the temperature rise of Comparative Example 1 reaches 18 degrees Celsius, Comparative Example 2 is 15 degrees Celsius, and Example 2 is controlled at 13 degrees Celsius. It can be concluded that Example 2, by establishing a three-dimensional temperature field model based on multi-point temperature data and calculating a comprehensive temperature compensation coefficient, effectively reduces the temperature rise during charging, averaging 8 to 12 degrees Celsius lower than Comparative Example 1. This is significant for extending battery life and improving charging safety. Figure 1The charging efficiency comparison curves show that charging efficiency initially increases and then decreases with temperature. At a low temperature of -10 degrees Celsius, the charging efficiency of Comparative Example 1 is only 76%, while Comparative Example 2 increases to 81%, and Example 2 reaches 83%. At 10 degrees Celsius, the efficiency of Comparative Example 1 rises to 85%, Comparative Example 2 to 88%, and Example 2 to 91%. In the optimal temperature range of 30 degrees Celsius, the efficiency of Comparative Example 1 is 88%, Comparative Example 2 is 90%, and Example 2 reaches 94%. At a high temperature of 50 degrees Celsius, the efficiency of Comparative Example 1 decreases to 84%, Comparative Example 2 to 87%, and Example 2 to 89%. It can be concluded that Example 2 has the highest charging efficiency at all temperature points, averaging 6% to 9% higher than Comparative Example 1. This is because Example 2 employs a comprehensive optimization strategy of operating condition identification, temperature compensation, and power prediction allocation, reducing energy loss. Figure 1 The SOC estimation error comparison curves show that accurate SOC estimation is fundamental to charging control. At -10 degrees Celsius, the SOC estimation error of Comparative Example 1 is 6.5%, Comparative Example 2 is 5.2%, and Example 2 reduces it to 2.8%. At 10 degrees Celsius, the error of Comparative Example 1 is 4.2%, Comparative Example 2 is 3.8%, and Example 2 is 1.9%. At 30 degrees Celsius, the error of Comparative Example 1 is 3.5%, Comparative Example 2 is 3.0%, and Example 2 is 1.6%. At 50 degrees Celsius, the error of Comparative Example 1 rebounds to 4.8%, Comparative Example 2 is 3.5%, and Example 2 is 2.1%. It can be concluded that Example 2, using a method that combines extended Kalman filtering and a long short-term memory neural network to jointly estimate SOC, achieves high-precision SOC estimation through Bayesian fusion, with the error controlled within 3% across the entire temperature range, significantly outperforming Comparative Example 1 and Comparative Example 2.

[0092] Experiment 2 simulated seven typical operating conditions of the work platform, including static standby, low-speed travel, high-speed travel, steering maneuver, light-load boom operation, heavy-load boom operation, and mixed operation. Comparative Examples 1 and 2 did not consider changes in operating conditions and used a fixed charging current of 25 amps in all conditions. In contrast, Example 2 dynamically adjusted the charging current based on the operating condition type identified by the operating condition recognition unit, achieving intelligent coordination between charging and operation through an adjustment coefficient. The experiment was conducted at an ambient temperature of 30 degrees Celsius, simulating various operating condition transitions in a real-world work scenario to evaluate the impact of different charging methods on the power supply for operation. Figure 2 The two subplots in the group show the comparative test results of charging strategies under different operating conditions in Experiment Example 2.

[0093] from Figure 2The bar chart comparing charging current adjustment strategies under different operating conditions shows that Example 2 achieves true adaptive charging control. In the static standby condition, the charging current for all three methods is 25 amps because the platform is unloaded and can charge at full power. In the low-speed travel condition, Comparative Examples 1 and 2 maintain a charging current of 25 amps, but Example 2 adjusts the current to 22.5 amps with an adjustment factor of 0.90, reserving power margin for travel. In the high-speed travel condition, Example 2 further reduces the current to 20 amps with an adjustment factor of 0.80. In the steering maneuver condition, the current in Example 2 drops to 18.5 amps with an adjustment factor of 0.74 because steering requires high power responsiveness. In the light-load boom operation condition, the current in Example 2 is 22 amps with an adjustment factor of 0.88. In the heavy-load boom operation condition, Example 2 significantly reduces the current to 16.3 amps with an adjustment factor of 0.65, ensuring sufficient power supply to the hydraulic system. Under mixed operating conditions, the current in Example 2 is 19.5 amps, and the adjustment factor is 0.78. It can be concluded that Example 2, through operating condition identification and adaptive current adjustment, solves the problem that traditional fixed-current charging methods cannot adapt to changes in operating conditions, achieving a dynamic balance between charging and operation. Figure 2 The bar chart comparing the performance of mixed operations over 50 cycles shows that Example 2 has a more significant advantage in actual operation scenarios. Comparative Example 1 accumulated a charging time of 32 hours in 50 consecutive charge-discharge cycles and experienced 15 power shortage alarms. These alarms were caused by excessive charging current consuming too much engine power, resulting in insufficient operating power. Comparative Example 2 accumulated a charging time of 30 hours and experienced 12 power shortage alarms. Although this was an improvement over Comparative Example 1, the problem still existed. Example 2 accumulated a charging time of only 27 hours and did not experience any power shortage alarms, achieving perfect coordination between charging and operation. It can be concluded that Example 2 reduced the charging time by 15.6% compared to Comparative Example 1 and by 10% compared to Comparative Example 2. More importantly, it completely eliminated power shortage alarms, improving operational continuity and reliability, and demonstrating the effectiveness of the operating condition identification and adaptive charging strategy.

[0094] Experiment 3 was designed according to the national standard GB / T31484-2015, "Cycle Life Requirements and Test Methods for Power Batteries for Electric Vehicles." Three battery packs underwent charge-discharge cycle testing at a constant temperature of 30 degrees Celsius. During the charging phase, each battery used its own charging method, and during the discharging phase, it was discharged at a constant current rate of 1C until the cutoff voltage. Capacity retention and internal resistance growth rate were tested every 50 cycles, for a total of 900 cycles. Capacity retention was calculated by dividing the measured capacity by the initial rated capacity of 50 amp-hours, and the internal resistance growth rate was measured using an AC internal resistance meter. The number of abnormal temperature alarms and the maximum voltage difference between individual cells were also recorded during the experiment as auxiliary indicators of battery health. Figure 3 The four subplots in the group show the long-term cycle life test results of Experiment Example 3.

[0095] from Figure 3 The comparison curves of capacity retention rate in the cycle life test show that the battery capacity gradually decreases with the increase of the number of cycles. After 900 cycles, the capacity retention rate of Comparative Example 1 dropped to 86.5%, which is close to the 80% SOH threshold. After 900 cycles, the capacity retention rate of Comparative Example 2 was 89.2%, and the decay rate was significantly slower than that of Comparative Example 1. After 900 cycles, the capacity retention rate of Example 2 remained at 93.2%, which is much higher than the other two groups. Based on the trend of capacity decay curves, extrapolation prediction shows that the number of cycles required for Comparative Example 1 to reach 80% SOH is approximately 1200 cycles, for Comparative Example 2 it is approximately 1420 cycles, and for Example 2 it is approximately 1560 cycles. It can be concluded that Example 2, through its strategy of adaptively adjusting charging parameters according to the SOH value, effectively slows down the battery aging rate, extending the life by 30% compared to Comparative Example 1 and by approximately 10% compared to Comparative Example 2. This is of great significance for reducing battery replacement costs and improving equipment economy. Figure 3 The comparison curves of internal resistance growth rates in the cycle life tests show that internal resistance is an important parameter reflecting the degree of battery aging. Comparative Example 1 shows a rapid increase in internal resistance in the early stages of cycling, reaching a growth rate of 16.1% in the first 450 cycles and increasing to 28% after 900 cycles. Comparative Example 2 shows a relatively slower growth rate, with a growth rate of 23% after 900 cycles. Example 2 shows the slowest internal resistance growth, exhibiting an almost linear and gradual growth trend, with a growth rate of only 15% after 900 cycles. It can be concluded that Example 2, through a comprehensive strategy of multi-stage adaptive charging, precise temperature compensation, and SOH adaptive parameter adjustment, effectively suppressed the growth of battery internal resistance and slowed down the battery aging process, which is consistent with the capacity retention results. Figure 3The bar chart comparing the overall performance after 900 cycles reveals that, in addition to capacity and internal resistance, battery consistency and safety are also important evaluation dimensions. Comparative Example 1 experienced 37 temperature anomaly alarms during 900 cycles, with a maximum single-cell voltage difference of 85 mV, indicating poor internal consistency within the battery pack. Comparative Example 2 experienced 18 temperature anomaly alarms, with a maximum single-cell voltage difference of 62 mV, showing improved consistency. Example 2 only experienced 5 temperature anomaly alarms, with a maximum single-cell voltage difference of only 35 mV, demonstrating excellent consistency. It can be concluded that the equalization charging strategy and multi-point temperature compensation mechanism of Example 2 effectively maintained the consistency of the battery pack, reduced temperature anomaly events, and improved long-term safety and reliability. Figure 3 The bar chart comparing the predicted cycle life at 80% SOH shows that, through mathematical modeling and extrapolation analysis of the capacity decay curve, the remaining battery life can be quantitatively predicted. Comparative Example 1 predicts a lifespan of 1200 cycles, Comparative Example 2 1420 cycles, and Example 2 1560 cycles. It can be concluded that the cycle life of Example 2 is 30% higher than Comparative Example 1 and approximately 9.9% higher than Comparative Example 2. This means that in practical applications, battery packs using the method of Example 2 can be used for a longer period, reducing replacement frequency and significantly lowering the total lifespan cost.

[0096] Experiment 4 simulated the actual working scenario of the work platform during an 8-hour work cycle, recording key parameters such as the cumulative running time of the diesel engine, diesel consumption, charging energy, and discharging energy. Comparative Example 1, Comparative Example 2, and Example 2 platforms performed the same work task. Diesel consumption was measured using a flow meter, and charging / discharging energy was measured using a power meter. The overall energy utilization efficiency was calculated as the ratio of output effective work to input fuel energy. The experiment also focused on the energy recovery function of Example 2, activating the energy recovery mode when the main telescopic boom descended and the vehicle decelerated, converting mechanical energy into electrical energy and storing it back in the battery pack. Figure 4 The four subplots in the group illustrate the comparative test results of energy utilization efficiency in Experiment 4. From... Figure 4 The bar chart comparing the cumulative operating times of the diesel engines shows that Comparative Example 1's engine operated for 7.2 hours within an 8-hour work cycle, accounting for 90% of the total operating time. Comparative Example 2's operating time was 6.8 hours, accounting for 85%. Example 2's operating time was only 6.3 hours, accounting for 78.8%. It can be concluded that Example 2, through power prediction and intelligent power source switching strategies, reduced unnecessary operating time of the diesel engine by 12.5% ​​compared to Comparative Example 1. This not only reduced fuel consumption but also reduced engine wear and maintenance requirements. Figure 4The bar chart comparing diesel consumption over an 8-hour operating cycle shows that fuel consumption is closely related to engine running time. Comparative Example 1 consumed 2.8 liters of diesel, Comparative Example 2 consumed 2.6 liters, and Example 2 consumed only 2.4 liters. It can be concluded that Example 2 reduces diesel consumption by 14.3% compared to Comparative Example 1 and by 7.7% compared to Comparative Example 2. Under the condition of large-scale operation throughout the year, this can save considerable fuel costs while reducing carbon emissions, meeting the requirements of energy conservation and environmental protection. Figure 4 The bar chart comparing charge and discharge energy shows that the balance between charging and discharging energy reflects the energy conversion efficiency. Comparative Example 1 has a charging energy of 1.28 kWh and a discharging energy of 1.12 kWh, with a charge / discharge efficiency of 87.5%. Comparative Example 2 has a charging energy of 1.31 kWh and a discharging energy of 1.18 kWh, with an efficiency of 90.1%. Example 2 has a charging energy of 1.36 kWh and a discharging energy of 1.27 kWh, with an efficiency of 93.4%. Example 2 also implemented an energy recovery function, accumulating 0.05 kWh of energy recovery during 12 activations of the energy recovery mode in the test period. It can be concluded that Example 2 has the highest charge / discharge efficiency, and the energy recovery function further improves energy utilization, demonstrating the advantages of an advanced energy management strategy. Figure 4 The comparative bar chart of comprehensive energy utilization efficiency shows that this is a comprehensive indicator for evaluating the performance of the entire energy system. Comparative Example 1 has a comprehensive efficiency of 45%, Comparative Example 2 has 48%, and Example 2 reaches 52%. It can be concluded that Example 2, through comprehensive measures such as predicting future power demand, intelligently switching between dual power sources, dynamically allocating charging power, and initiating energy recovery mode, improved the comprehensive energy utilization efficiency by 7 percentage points, representing a 15.6% improvement compared to Comparative Example 1 and an 8.3% improvement compared to Comparative Example 2, significantly optimizing energy utilization.

[0097] Experiment 5 was conducted on an inclined platform and a vibration table to simulate various extreme working conditions that the platform might encounter in actual operation. For the tilt test, the battery pack was mounted on an adjustable-angle platform with five tilt angles: 0 degrees, 3 degrees, 5 degrees, 8 degrees, and 10 degrees. Charging tests were performed at each angle, and the charging current, individual cell voltage difference, and temperature distribution were recorded. For the vibration test, four vibration acceleration levels of 0.3g, 0.5g, 0.8g, and 1.0g were applied to the vibration table, and the response of the charging system was observed. Comparative Examples 1 and 2 did not have tilt and vibration protection functions, while Example 2 was equipped with a complete safety protection unit capable of automatically adjusting or stopping charging based on the platform's attitude and vibration status. Figure 5 The four sub-graphs of the group illustrate the test results of the safety protection function verification in Experiment Example 5. From Figure 5The comparison curves of charging current adjustment under different tilt angles show that Example 2 achieves tilt adaptive protection. Comparative Examples 1 and 2 maintain a constant charging current of 25 amps at all tilt angles without any protection measures. Example 2 maintains normal charging at 25 amps when the tilt angle is less than 3 degrees, reduces the current to 22.5 amps when the tilt angle reaches 3 degrees, further reduces it to 17.5 amps when the tilt angle is 5 degrees, and completely stops charging and triggers an audible and visual alarm when the tilt angle is 8 degrees or higher. It can be concluded that the tilt protection strategy of Example 2 can adjust the charging current in stages according to the degree of platform tilt, and stop charging in time during severe tilting, effectively avoiding safety hazards under tilted conditions. Figure 5 The bar chart comparing safety indicators after charging at an 8-degree tilt for 30 minutes shows that continuous charging under tilt conditions leads to serious safety issues. In Comparative Example 1, after charging at an 8-degree tilt for 30 minutes, the individual cell voltage difference increased to 120 millivolts, and the maximum temperature difference reached 8 degrees Celsius, indicating significant unevenness within the battery pack and a risk of localized overheating. Comparative Example 2 showed a slightly better situation with a 95-mV individual cell voltage difference and a maximum temperature difference of 6 degrees Celsius, still posing a safety hazard. Example 2 immediately stopped charging at an 8-degree tilt, resulting in zero individual cell voltage and temperature differences, completely avoiding safety issues. It can be concluded that the tilt protection function in Example 2 is necessary and effective, capable of promptly cutting off charging under abnormal tilt conditions to prevent internal imbalance and overheating of the battery pack. Figure 5 The comparison curves of charging current adjustment under different vibration intensities show that Example 2 also has a vibration adaptive protection function. Comparative Examples 1 and 2 maintained a charging current of 25 amps under all vibration acceleration conditions without any adjustment. Example 2 maintained normal charging at 25 amps when the vibration acceleration was 0.3g. When the vibration reached 0.5g and lasted for more than 3 seconds, the charging current was significantly reduced to 8.75 amps, which is 35% of the original current, greatly reducing the heat and stress caused by charging. When the vibration reached 0.8g or higher, Example 2 immediately stopped charging and resumed charging after a 5-second delay when the vibration returned to normal. Under severe vibration conditions of 1.0g, Comparative Examples 1 and 2 showed poor contact of the charging cable, with voltage fluctuation reaching 8%, indicating serious connection problems and safety hazards. However, Example 2 had stopped charging and was therefore unaffected. It can be concluded that the vibration protection strategy of Example 2 can effectively cope with vibration conditions, avoiding vibration-induced loose connections, voltage fluctuations, and potential safety accidents by reducing or suspending charging. Figure 5The comparative bar chart of various safety incidents shows that the effectiveness of the safety protection function is reflected in the reduction of safety incidents throughout the testing cycle. Comparative Example 1 experienced 37 temperature anomalies, 15 power alarms, and 5 other anomalies during the test, totaling 57 safety incidents. Comparative Example 2 experienced 18 temperature anomalies, 12 power alarms, and 3 other anomalies, totaling 33. Example 2 experienced only 5 temperature anomalies, 0 power alarms, and 0 other anomalies, totaling 5, representing 8.8% of the number of safety incidents in Comparative Example 1. It can be concluded that Example 2, through the comprehensive effect of multiple safety protection mechanisms such as tilt monitoring, vibration monitoring, and connection monitoring, reduced safety incidents by 91.2%, significantly improving the safety of the charging process and the reliability of the system.

[0098] The comparative verification through the above experiments demonstrates that the battery pack charging control method for the narrow-body self-propelled work platform of the present invention has significant advantages over traditional methods. The method in Example 2 reduces charging time by 15% to 25%, increases charging efficiency by 6% to 9%, reduces battery temperature rise by 8 to 12 degrees Celsius, reduces SOC estimation error by less than 3%, extends battery cycle life by 20% to 30%, improves energy utilization efficiency by 7 percentage points, reduces diesel consumption by 14.3%, and enhances safety protection functions. Experimental data can be visually displayed using MATLAB software to plot charging curves, temperature curves, SOC estimation accuracy curves, capacity decay curves, etc., fully verifying the beneficial effects of the present invention.

[0099] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A battery pack charging control method for a narrow-body self-propelled work platform, characterized in that, Includes the following steps: S1: Collect operating parameters of the lithium battery pack and operating parameters of the work platform; S2: Construct a working condition feature vector to identify the current working condition type; S3: Select the charging stage and determine the reference charging current based on the operating condition type and SOC value; S4: Establish a three-dimensional temperature field model based on multi-point temperature data, calculate the comprehensive temperature compensation coefficient, and adjust the charging current; S5: The extended Kalman filter and long short-term memory neural network are used to jointly estimate SOC and SOH and correct the charging parameters; S6: Predicts future power demand, switches between the power source of the diesel engine and the electric pump, and dynamically allocates charging power; S7: Detects platform tilt angle and vibration acceleration, and adjusts or stops charging accordingly; S8: Uploads charging and discharging data to the cloud server, receives optimized charging strategy parameters, and updates the local algorithm.

2. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S1 includes the following steps: S11: The voltage of each individual cell in the battery pack is acquired through the voltage acquisition module at a sampling frequency of 100Hz to 1000Hz, with a voltage acquisition accuracy of ±5mV to ±10mV. S12: The charging current or discharging current is acquired by a current sensor connected in series in the main charging and discharging circuit at a sampling frequency of 500Hz to 2000Hz. The current sensor measurement accuracy is ±0.3% to ±0.8%. S13: Collect the temperature at multiple points inside the battery pack using 12 to 20 temperature sensors, which are distributed at the positive terminal, negative terminal, middle of the battery pack, and outer surface. S14: The walking speed is measured by the walking speed sensor installed on the walking wheel, the steering angle is measured and the steering angular velocity is calculated by the steering angular displacement sensor installed on the steering mechanism, and the hydraulic system power is calculated by the pressure sensor and flow sensor installed on the hydraulic system oil pipe. S15: The tilt angle and vibration acceleration of the platform in the X, Y, and Z directions are measured by a six-axis attitude sensor installed at the center of the walking platform. S16: All collected parameters are transmitted to the main controller via the CAN bus. The main controller performs validity verification and filtering on the data.

3. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S2 includes the following steps: S21: Constructing a seven-dimensional feature vector for operating conditions: ； in, Walking speed; This refers to the steering angular velocity; Power of the hydraulic system; The platform tilt angle; It is the vibration acceleration; This is the current driving the motor; This refers to the boom load power; S22: Normalize the seven-dimensional working condition feature vector. The normalization formula is as follows: ; in, is the normalized value of the i-th feature; This represents the actual measured value of the i-th feature; The minimum value of the i-th feature; The maximum value of the i-th feature; S23: Input the normalized feature vector into a pre-trained fuzzy neural network, which includes 7 input nodes, 15 to 25 hidden layer nodes and 7 output layer nodes; S24: Calculate the output value of each node in the output layer, select the working condition type corresponding to the node with the largest output value as the identification result, and identify it as one of the following: static standby working condition, low-speed walking working condition, high-speed walking working condition, steering maneuvering working condition, boom light load action working condition, boom heavy load action working condition, or mixed operation working condition.

4. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S3 includes the following steps: S31: Determine the current SOC value range. When SOC is less than 0.20, select the pre-charging stage; when SOC is between 0.20 and 0.60, select the fast charging stage; when SOC is between 0.60 and 0.85, select the optimized charging stage; when SOC is between 0.85 and 0.95, select the equalization charging stage; when SOC is greater than or equal to 0.95, select the float charging maintenance stage. S32: Determine the reference charging current based on the selected charging stage and the current operating condition. During the pre-charging phase, settings are configured for stationary standby, low-speed travel, and high-speed travel conditions. The setting is 0.15C to 0.25C, for steering maneuvering and boom operation conditions. The voltage range is 0.03C to 0.07C; during the fast charging phase, the voltage is set under standby conditions. The current is set to 0.45C to 0.55C. During the walking operation, pulse charging is used and the average current is calculated. During the boom operation, the current is set... The temperature ranges from 0.25°C to 0.35°C. S33: During the optimized charging phase, according to the formula... Calculate the reference charging current, where The maximum charging current is set between 0.4C and 0.5C; during the equalization charging phase, the voltage difference between individual cells is monitored. According to the formula Calculate the reference charging current, where This is the difference between the highest and lowest voltage of a single battery cell; during the float charging maintenance phase, constant voltage charging is used, with the charging voltage set to 54.4V to 54.8V; S34: Determine the operating condition adjustment coefficient based on the operating condition type. In standby mode Low-speed travel conditions Up to 0.95, under high-speed driving conditions Up to 0.85, under steering maneuvering conditions Up to 0.80, under light load operating conditions of the boom Up to 0.90, under heavy-load boom operation conditions Up to 0.70, under mixed operating conditions Up to 0.75; S35: Calculate the adjusted charging current ;in This is the adjusted charging current.

5. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S4 includes the following steps: S41: Read the real-time temperature data from all temperature sensors to obtain the temperature vector. ,in Let be the measured temperature of the i-th temperature sensor; S42: Determine the position weighting coefficient based on the location of each temperature sensor. The weighting coefficient at the positive and negative terminals is 0.15 to 0.20, the weighting coefficient at the middle of the battery pack is 0.10 to 0.15, and the weighting coefficient at the surface of the battery pack casing is 0.05 to 0.

10. S43: Calculate the single-point temperature compensation factor for each temperature measurement point. : when When the temperature is below -10℃ Up to 0.6, when Between -10℃ and 0℃ to ; when Between 0℃ and 20℃ to ; when Between 20℃ and 35℃ Up to 1.05; when Between 35℃ and 50℃ to ; when When the temperature is above 50℃ Up to 0.5; S44: Calculate the overall temperature compensation coefficient using the weighted average method: ； in This is the comprehensive temperature compensation coefficient; Σ represents the position weighting coefficient of the i-th temperature measurement point; Σ represents the summation over all temperature measurement points. S45: Calculate the maximum temperature gradient within the battery pack: ； in The maximum temperature gradient; The absolute value of the temperature difference between any two temperature measurement points; The distance between two temperature measuring points; when When it exceeds 5℃ / dm, Further reduced to to At the same time, the air-cooling system is activated; S46: Correct the charging current based on the comprehensive temperature compensation coefficient, and calculate the temperature-compensated charging current. ;in This is the charging current after temperature compensation.

6. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S5 includes the following steps: S51: Establish a second-order RC equivalent circuit model and define the state vector: ;in Let k be the state of charge at time k; and These represent the voltages of the two RC branches; T denotes matrix transpose. S52: Execute the extended Kalman filter algorithm to calculate the prior state estimate: ;in For prior state estimation; A represents the posterior state estimate from the previous time step; A is the state transition matrix; B is the input matrix. This represents the charging and discharging current at the previous moment; S53: Calculate the Kalman gain and update the posterior state estimate, extracting from the posterior state estimate... As an estimate of the SOC in the extended Kalman filter algorithm; S54: Construct a feature vector containing 30 consecutive sampling points. The feature vector includes SOC, voltage, current, temperature, internal resistance, voltage change after resting, and cumulative loop count. Input the feature vector into a trained Long Short-Term Memory (LSTM) neural network, which includes a first LSTM layer, a Dropout layer, a second LSTM layer, and a fully connected layer. Output... and As estimates of SOC and SOH for neural networks; S55: Calculate the confidence variance of the extended Kalman filter and neural network estimations, and calculate the weighting coefficients based on the confidence variance: and ;in The confidence variance estimated by extended Kalman filtering; The confidence variance estimated for the neural network; S56: Calculate the final SOC and SOH values ​​using Bayesian fusion: ； in This is a correction factor, ranging from -0.8 to -1.2; This represents the relative change in internal resistance. S57: Correct charging parameters based on the final SOH value, when When the current is less than 0.90, the maximum charging current will be reduced to 0.35C to 0.45C. When the value is less than 0.85, the maximum charging current is further reduced to 0.30C to 0.40C. When the voltage is less than 0.80, the charging cutoff voltage will be reduced by 0.03V to 0.08V.

7. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S6 includes the following steps: S61: Collect historical power data from the past 5 to 10 minutes, establish a power series, and use a sliding window time series analysis method to predict the power demand for the next 30 to 60 seconds. The prediction formula is: ； in Δt represents the predicted power; Δt represents the prediction duration. Historical data weighting coefficients; Historical power values; This represents the power change trend term; S62: Determine the power source switching conditions. When the load power is greater than 0.80 to 0.90 times the maximum power of the electric pump, or the SOC is less than 0.25 to 0.35, or the predicted power is greater than 0.85 to 0.95 times the rated power of the electric pump, switch from the electric pump to the diesel engine. When the load power is less than 0.55 to 0.65 times the output power of the diesel engine, and the SOC is less than 0.85 to 0.95, and the duration is greater than 8s to 12s, switch from the diesel engine to the charging mode. S63: In charging mode, calculate the remaining power available for charging: ； in The remaining power available for charging; This refers to the output power of the diesel engine. For load power; Reserved power determined based on predicted power; S64: Calculate charging power: ； in This represents the actual charging power; min indicates taking the minimum value. This is the maximum charging power; For charging efficiency, the value ranges from 0.88 to 0.95; This is the power adjustment factor based on SOC; This is a temperature-based power adjustment coefficient; S65: Calculate charging current based on charging power and battery voltage. ; in This refers to the charging current calculated based on power allocation. This is the current voltage of the battery pack; S66: When the boom is lowered or the vehicle is decelerated, detect whether the drive motor is in the generator state. When it is in the generator state and the energy recovery conditions are met, start the energy recovery mode, calculate the energy recovery current and limit it to the range of 0.3C to 0.6C.

8. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S7 includes the following steps: S71: Read the platform's forward / backward tilt angle α and left / right tilt angle β from the six-axis attitude sensor, and calculate the overall tilt angle. ;in For overall tilt; S72: Adjust charging current based on overall tilt angle: when The charging current remains constant when the temperature is less than 3°. when The charging current is adjusted when the temperature is between 3° and 5°. to ; when The charging current is adjusted when the temperature is between 5° and 8°. to ; when Charging will stop and an audible and visual alarm will be triggered when the temperature is greater than or equal to 8°C. S73: Reads three-axis acceleration data from a six-axis attitude sensor. , and Calculate the overall vibration acceleration ,in For comprehensive vibration acceleration; S74: Adjust the charging current based on the overall vibration acceleration. when If the amount exceeds 0.4g to 0.6g and lasts for more than 2s to 4s, reduce the charging current to [a lower value]. to ; when If the amount exceeds 0.8g to 1.2g, charging will be paused immediately. Charging will resume 3 to 7 seconds after the vibration returns to normal. S75: Monitors voltage fluctuations in the charging circuit and calculates the voltage fluctuation rate. ;in Voltage fluctuation rate; The maximum voltage within the sampling window; Minimum voltage; For average voltage; when If the current exceeds 0.04 to 0.06, it is determined that the connector may be loose, the charging current is reduced to 0.10C to 0.20C and an alarm is triggered; S76: Select the smaller value between the tilt-adjusted charging current and the vibration-adjusted charging current as the final safe charging current, and send it to the charging controller for execution.

9. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 1, characterized in that, S8 includes the following steps: S81: After each operation, read charge / discharge curve data, historical temperature data, estimated SOC and SOH values, operating condition type statistics, abnormal event records and environmental parameters from the data storage module; S82: Package the data into a data packet in JSON or XML format, the data packet containing vehicle identification number, operation date, battery data, operating condition data, abnormal events and environmental data fields; S83: Uploads data packets to the cloud server using the HTTPS protocol via a 4G / 5G wireless communication module or WiFi module; S84: The cloud server receives data from multiple operating platforms and uses a big data processing framework to perform cluster analysis of similar working conditions, evaluate the effectiveness of charging strategies, train battery life prediction models, and perform statistical analysis of optimal charging parameters. S85: The cloud server generates optimized charging strategy parameters based on big data analysis results. The parameters include the current setting value, charging cut-off voltage, equalization charging time, temperature compensation coefficient matrix, and operating condition adjustment coefficient for each charging stage. S86: The cloud server packages the optimized charging strategy parameters into a configuration file and pushes it to the operation platform; S87: The main controller of the operating platform receives the configuration file, performs integrity verification and version compatibility checks, and then updates the parameters of the charging stage selection unit, temperature compensation processing unit, and energy management unit. S88: The cloud server predicts the remaining battery life (RUL) and failure probability based on the battery aging trend. When the RUL is less than the preset threshold or the failure probability is greater than the preset threshold, an early warning message is generated and pushed to the operation platform and management platform.

10. The battery pack charging control method for a narrow-body self-propelled work platform according to claim 4, characterized in that, In step S32, during the fast charging phase while the vehicle is in motion, a pulse charging mode is used, and the duty cycle of the pulse charging is... ;in Duty cycle; Current walking power; Maximum walking power; pulse period is 2s to 5s, current is 0.5C during charging pulse, current is 0.05C to 0.10C during discharging pulse, and charging pulse duration is... The discharge pulse duration is ;in The pulse period; This refers to charging time; This refers to the discharge time.

Images