Forklift Power Battery Pack Management Method, Device, Equipment and Storage Medium

Abstract

The present invention relates to the technical field of forklift power battery pack management, and discloses a forklift power battery pack management method, device, equipment, storage medium and system, including a dynamic load prediction module, a multi-mode thermal management module and an adaptive current control module. Through the deep coupling of the dynamic load prediction model optimized by edge computing and the multi-mode thermal management system, a real-time collaborative closed loop is formed, solving the problem of the lag in the response of load changes and battery temperature rise. The closed-loop optimization of the data flow between modules realizes adaptive adjustment under multiple working conditions, and significantly improves the battery life and energy efficiency under the hardware conditions with limited resources. The mechanism of coupling and adjusting the experimentally calibrated cooling coefficient and the real-time load effectively balances the cooling energy consumption and the temperature control requirements, and significantly reduces the system energy consumption. At the same time, the long-term data recording and optimization iteration supported by the cloud platform enhance the intelligent adaptability of the system, enabling the present invention to have the ability to continuously optimize its performance in complex operating environments.

Description

Technical Field

[0001] The present invention relates to a management method, device, equipment, storage medium and system for a forklift power battery pack, and belongs to the technical field of vehicle battery management. Background Art

[0002] Forklift power battery packs are widely used in fields such as logistics transportation and warehousing management. Their core function is to provide stable power support for forklifts under complex working conditions such as heavy loads and high-frequency starting and stopping. In the prior art, the management of forklift power battery packs usually relies on traditional load prediction and static thermal management modes, aiming to extend battery life and improve energy utilization efficiency. However, these technologies still face many problems in practical applications.

[0003] To improve load adaptability, traditional methods use load prediction models based on fixed working conditions to guide the discharge strategy of the battery pack. However, these models lack real-time and dynamic adjustment capabilities and cannot respond to load changes in a timely manner under rapidly changing working conditions (such as high-frequency starting and stopping or heavy-load operations), resulting in problems such as battery overload or uneven discharge. To solve this problem, the industry usually increases redundant battery capacity or designs complex control algorithms, but this inevitably increases hardware costs and system complexity.

[0004] In terms of thermal management, the prior art mostly uses passive heat dissipation or simple air-cooling and liquid-cooling methods to adjust the cooling power according to static heat dissipation modes. However, this method is difficult to meet the dynamic thermal management requirements of complex working conditions and is prone to problems such as heat dissipation lag or excessive energy consumption. To overcome this defect, the industry attempts to optimize heat dissipation through preset cooling strategies based on historical working conditions. However, this preset method is insufficient in responding to sudden high loads and ambient temperature changes and may still lead to situations such as battery overheating or insufficient cooling. For example: 1. Traditional load prediction models lack real-time dynamic adjustment capabilities and cannot meet the load change requirements under complex working conditions. 2. Existing thermal management technologies have a lag in response, high energy consumption, and are difficult to achieve precise heat dissipation under dynamic working conditions. 3. The cooperation between modules is insufficient, and data isolation results in limited system optimization effects. Therefore, how to achieve the collaborative optimization of dynamic load prediction and multi-mode thermal management systems and significantly improve the comprehensive performance of forklift power battery packs has become the technical problem to be solved by the present invention. Summary of the Invention

[0005] The present invention provides a management method, device, equipment and storage medium for a forklift power battery pack, and its main purpose is to solve the problems of lag in dynamic load response, low thermal management efficiency and insufficient cooperation between modules.

[0006] To achieve the above object, a management method for a forklift power battery pack provided by the present invention includes the following management steps: Step 1, collect data in real time through a data acquisition module: collect the voltage, current, temperature of the battery pack and the operating load data of the forklift in real time through a distributed sensor module, where: the voltage data is used to evaluate the state of the battery unit; the current data is used to calculate the instantaneous load demand of the forklift; the temperature data is used to monitor the thermal stability of the battery pack; the operating load data is dynamically recorded based on the operating mode of the forklift.

[0007] Step 2, perform dynamic load prediction through a dynamic load prediction module: based on the long short-term memory network LSTM algorithm, establish a prediction model, and perform segmented processing on the real-time collected load data; use historical load data to train the model weights, and update the model parameters in real time to optimize the prediction accuracy; predict the load demand in the next time period, specifically quantify the amplitude and frequency of the load fluctuation, and generate a load prediction value , and the calculation formula is: ; where and are the historical data weight and the real-time data weight respectively, is the hidden state value at the previous moment, is the current input data, is the bias term, is the activation function;

[0008] Step 3, perform multi-mode thermal management through a multi-mode thermal management module: according to the real-time temperature data and the load prediction value, dynamically select a cooling mode: when the temperature is less than 30 °C, activate the passive heat dissipation mode; when the temperature is between 30 °C and 45 °C, activate the air-cooling mode, and the cooling power is adjusted according to the formula: ; where is the cooling power adjustment coefficient; when the temperature exceeds 45 °C, activate the liquid-cooling mode, and dynamically adjust the cooling intensity according to the load prediction value : ; where is the load-related cooling coefficient;

[0009] Step 4, perform adaptive current control through an adaptive current control module: according to the load prediction value and the battery pack state parameters, adjust the discharge current of each battery unit in real time to balance the load distribution among the battery units. The specific control rule is: ; where is the number of battery units, is the unit cell voltage; limit the current change rate of the unit cell to ensure that the change amplitude per second does not exceed 5% of the rated value;

[0010] Step 5, inter-module collaborative optimization: The dynamic load prediction module and the multi-mode thermal management module work together to adjust the cooling mode in real time according to the load prediction value; The adaptive current control module dynamically optimizes the battery cell discharge strategy based on the load prediction and thermal management feedback; Data is transmitted and processed through a unified edge computing device;

[0011] Step 6, data feedback and system update: Data is transmitted to the cloud platform through the edge computing device, and the cloud analyzes the performance of the prediction model and the thermal management effect, and regularly updates the model weights and cooling strategy parameters; The cloud platform generates a dynamic maintenance plan, including battery maintenance suggestions and cooling system optimization guidance.

[0012] Preferably, the long short-term memory network (LSTM) model in the dynamic load prediction further includes: an embedding layer for dimensionality reduction processing of the real-time collected load data; a bidirectional LSTM layer for simultaneously capturing the forward and backward time dependencies of the load data; a dense connection layer for converting the output time series prediction value into the next moment load prediction value.

[0013] Preferably, the weight coefficient of the load prediction model and The initial value is determined by the following method: Normalize the historical load data; Use the least squares method to fit the prediction error and calculate the gradient of each iteration; Optimize the weight coefficient based on the gradient descent algorithm until the prediction error converges to a preset threshold.

[0014] Preferably, the cooling power adjustment coefficient of the multi-mode thermal management system and the load-related cooling coefficient are obtained through an experimental calibration method, which specifically includes: Measuring the heat loss at the cooling mode switching point under different temperature conditions; Calculating the ratio of energy consumption to heat dissipation in each cooling mode and optimizing to obtain the cooling coefficient.

[0015] Preferably, the adjustment strategy of the discharge current of the adaptive current control module further includes: Real-time correction of the discharge current according to the change in the internal impedance of the battery cell; The change in internal impedance is calculated by real-time collecting the voltage and current of the battery cell, and the formula is: ; where is the load terminal voltage.

[0016] Preferably, the collaborative optimization between the modules further includes the following steps: the dynamic load prediction module updates the prediction value every 1 second and transmits the prediction result to the multi-mode thermal management system; the multi-mode thermal management system updates the cooling strategy according to the load prediction value and feeds back the actual cooling efficiency to the adaptive current control module; the adaptive current control module adjusts the discharge current in real time and returns the adjustment result to the dynamic load prediction module to update the load model at the next moment.

[0017] Preferably, in the feedback update step, the following data is recorded by the cloud platform for model optimization: the actual error of each load prediction; the energy consumption and heat dissipation efficiency under each cooling mode; the discharge curve and life trend of the battery unit.

[0018] An apparatus for a forklift power battery pack management method, comprising: a data acquisition module for real-time acquisition of the voltage, current, temperature of the battery pack and the forklift operation load data; a prediction module for predicting the load demand at the next moment based on the long short-term memory network (LSTM) model; a thermal management module for dynamically adjusting the cooling mode according to the load prediction value and the temperature data; an current control module for real-time adjustment of the discharge current of the battery unit; an edge computing module for realizing data transfer and collaborative optimization between the modules.

[0019] An equipment for a forklift power battery pack management method, the equipment comprising: at least one apparatus, the apparatus including a data acquisition module, a prediction module, a thermal management module, an current control module and an edge computing module; a communication module for communicating with the cloud platform to realize data upload and feedback download; a storage module for storing load data, temperature data and model optimization parameters.

[0020] A storage medium for a forklift power battery pack management method, the storage medium storing computer-readable instructions, when the instructions are executed by a computing device, causing the computing device to implement the following steps: collecting real-time data of the battery pack and inputting it into the long short-term memory network model; adjusting the cooling strategy and the discharge current according to the load prediction value output by the model; transmitting the data to the cloud platform to optimize the model parameters; updating the operation parameters of the apparatus according to the feedback of the cloud platform.

[0021] Compared with the problems described in the background art, the beneficial effects of the present invention are as follows: Through a lightweight dynamic load prediction model optimized by edge computing, the real-time load prediction is deeply coupled with a multi-mode thermal management system, breaking through the limitations of single-module optimization, realizing two-way adjustment of the dynamic association between load changes and temperature rise, not only improving the prediction accuracy of the system, but also significantly reducing the response lag of the cooling system, enabling the load prediction and thermal management to form a real-time cooperation closed-loop; The system integrates a dynamic load prediction, an adaptive current control, and a multi-mode thermal management module. Through the closed-loop optimization of the real-time data flow between the modules, under the hardware conditions with limited resources, the adaptive optimization of multi-condition scenarios (such as high-frequency start-stop, heavy-load transportation, low-temperature operation) is achieved. This collaborative optimization strategy effectively balances the complex contradictions among battery life, discharge efficiency, and system energy consumption, demonstrating a comprehensive balance of multi-dimensional performance improvement; By adopting a dynamic adjustment mechanism that couples the experimentally calibrated cooling coefficient with real-time load prediction, the balance between cooling energy consumption and battery life is optimized. Especially in the case of frequent heavy loads or low-temperature conditions, this mechanism can reduce overheating or low-temperature losses through precise thermal management strategies, providing technical support for extending the battery life, while reducing additional energy consumption; Through the long-term data recording and analysis of the cloud platform, the system can continuously optimize the prediction model weights and thermal management strategy parameters based on the actual operation data, enabling the entire management method to have the ability to continuously improve performance in multi-scenarios and dynamic working conditions. This function significantly enhances the stability and intelligence of the system in complex operating environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a flowchart of the dynamic load prediction and thermal management system of the present invention.

[0023] Figure 2 It is a data interaction and working flowchart between each module in the forklift power battery pack management system of the present invention.

[0024] The implementation, functional features, and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0026] See Figure 1 and Figure 2 The embodiments of the present application provide a forklift power battery pack management method, device, equipment, and storage medium. The following management steps are included:

[0027] Step 1, collect data in real time through the data acquisition module: The voltage, current, temperature of the battery pack and the operating load data of the forklift are collected in real time through the distributed sensor module, where: the voltage data is used to evaluate the state of the battery unit; the current data is used to calculate the instantaneous load demand of the forklift; the temperature data is used to monitor the thermal stability of the battery pack; the operating load data is dynamically recorded based on the operating mode of the forklift.

[0028] Step 2, perform dynamic load prediction through the dynamic load prediction module: Based on the long short-term memory network (LSTM) algorithm, establish a prediction model, and segment the real-time collected load data; use historical load data to train the model weights, and update the model parameters in real time to optimize the prediction accuracy; predict the load demand in the next time period, specifically quantify the amplitude and frequency of the load fluctuation, and generate a load prediction value , and the calculation formula is: ; where and are the historical data weight and the real-time data weight respectively, , is the hidden state value at the previous moment, is the current input data, is the bias term, is the activation function;

[0029] Among them, the hidden state value at the previous moment ( ) is the output state of the LSTM model, which is used to store the correlation information between the predicted value at the current moment and the historical data. Its calculation formula is: ; where is the hidden state weight, is the input weight, is the input data at the current moment, is the bias term, is the activation function (such as tanh). For example, in the load prediction of the forklift power battery, the hidden state value is used to record the trend of load change at the previous moment, and guide the selection of the subsequent cooling mode and the optimization of the discharge path.

[0030] Step 3, perform multi-mode thermal management through the multi-mode thermal management module: According to the real-time temperature data and the load prediction value, dynamically select the cooling mode: when the temperature is less than 30°C, activate the passive heat dissipation mode; when the temperature is between 30°C and 45°C, activate the air cooling mode, and the cooling power is adjusted according to the formula: ; where is the cooling power adjustment coefficient; when the temperature exceeds 45°C, activate the liquid cooling mode, and dynamically adjust the cooling intensity according to the load prediction value : ; where is the load-related cooling coefficient; the multi-mode thermal management module provides real-time feedback on the cooling efficiency and adjusts the cooling strategy to avoid energy waste;

[0031] The cooling efficiency ( ) is calculated based on the cooling energy consumption and the temperature change, and the formula is: ; where is the temperature difference before and after cooling, is the energy consumed in the cooling mode. The cooling efficiency is monitored in real time by the sensor module for temperature and power data, and transmitted to the thermal management module through the edge computing module. The path for adjusting the cooling strategy is as follows: when the cooling efficiency is lower than the preset threshold (e.g., ), the thermal management module switches to a more efficient cooling mode (such as from air cooling to liquid cooling) and adjusts the cooling intensity according to the real-time load prediction value: ; where is the cooling coefficient, is the load prediction value.

[0032] Step 4, perform adaptive current control through the adaptive current control module: According to the load prediction value and the battery pack status parameters, the discharge current of each battery cell is adjusted in real time to balance the load distribution among the battery cells. The specific control rule is: ; where is the number of battery cells, is the voltage of the unit battery; limit the current change rate of the unit battery to ensure that the change amplitude per second does not exceed 5% of the rated value; the rated value is the designed discharge capacity of the battery cell, provided by the manufacturer. The formula for the current change rate is as follows: ; where seconds. Monitor the current value of the unit battery in real time. If the current change rate exceeds the limit value ( ), dynamically optimize the discharge current distribution by adjusting the output value of the load prediction module to ensure the operating safety of each battery cell.

[0033] Step 5, collaborative optimization among modules: The dynamic load prediction module and the multi-mode thermal management module work together to adjust the cooling mode in real time according to the load prediction value; the adaptive current control module dynamically optimizes the discharge strategy of the battery cells based on the load prediction and the thermal management feedback; the data is transmitted and processed through a unified edge computing device to achieve collaborative optimization among the modules; among them, the dynamic optimization of the discharge strategy of the battery cells is based on the real-time load prediction value , the battery cell status parameters (such as voltage , internal resistance ) and the cooling efficiency feedback from the thermal management module. The formula for the discharge current is: ; and correct according to the internal resistance change: ; The edge computing device is responsible for aggregating sensor data, running short-term prediction models, transmitting control instructions, and achieving collaborative optimization between modules through a unified communication protocol. The collaborative optimization between modules includes real-time updates of the dynamic load prediction module, adjustment of the cooling strategy of the thermal management module, and optimization of the discharge current of the adaptive current control module to ensure the real-time performance and efficiency of the system.

[0034] Step 6, data feedback and system update: The data is transmitted to the cloud platform through the edge computing device. The cloud analyzes the performance of the prediction model and the thermal management effect, and regularly updates the model weights and cooling strategy parameters; the cloud platform generates a dynamic maintenance plan, including battery maintenance suggestions and cooling system optimization guidance. Preferably, the long short-term memory network LSTM model in the dynamic load prediction further includes: an embedding layer for dimensionality reduction processing of the real-time collected load data; a bidirectional LSTM layer for simultaneously capturing the forward and backward time dependencies of the load data; a dense connection layer for converting the output time series prediction value into the next moment load prediction value.

[0035] Preferably, the weight coefficient of the load prediction model and The initial values of are determined by the following method: normalize the historical load data; use the least squares method to fit the prediction error and calculate the gradient of each iteration; optimize the weight coefficient based on the gradient descent algorithm until the prediction error converges to a preset threshold. Preferably, the cooling power adjustment coefficient of the multi-mode thermal management system and the load-related cooling coefficient are obtained through an experimental calibration method, which specifically includes: measuring the heat loss at the cooling mode switching point under different temperature conditions; calculating the ratio of energy consumption to heat dissipation in each cooling mode and optimizing to obtain the cooling coefficient.

[0036] Preferably, the adjustment strategy of the discharge current of the adaptive current control module further includes: correcting the discharge current in real time according to the change of the internal impedance of the battery cell; the change of the internal impedance is calculated by real-time collecting the voltage and current of the battery cell, and the formula is: ; where is the load terminal voltage.

[0037] Preferably, the collaborative optimization between the modules further includes the following steps: The dynamic load prediction module updates the prediction value every 1 second and transmits the prediction result to the multi-mode thermal management system; The multi-mode thermal management system updates the cooling strategy according to the load prediction value and feeds back the actual cooling efficiency to the adaptive current control module; The adaptive current control module adjusts the discharge current in real time and returns the adjustment result to the dynamic load prediction module to update the load model for the next moment.

[0038] Preferably, in the feedback update step, the following data is recorded by the cloud platform for model optimization: the actual error of each load prediction; the energy consumption and heat dissipation efficiency under each cooling mode; the discharge curve and life trend of the battery unit.

[0039] An apparatus based on a forklift power battery pack management method, comprising: a data acquisition module for real-time acquisition of the voltage, current, temperature of the battery pack and the forklift operation load data; a prediction module for predicting the load demand for the next moment based on the long short-term memory network (LSTM) model; a thermal management module for dynamically adjusting the cooling mode according to the load prediction value and temperature data; an current control module for real-time adjusting the discharge current of the battery unit; an edge computing module for realizing data transfer and collaborative optimization between the modules.

[0040] A device based on a forklift power battery pack management method, the device comprising: at least one apparatus, the apparatus including a data acquisition module, a prediction module, a thermal management module, an current control module and an edge computing module; a communication module for communicating with the cloud platform to realize data upload and feedback download; a storage module for storing load data, temperature data and model optimization parameters.

[0041] A storage medium based on a forklift power battery pack management method, the storage medium storing computer-readable instructions, when the instructions are executed by a computing device, causing the computing device to implement the following steps: acquiring real-time data of the battery pack and inputting it into the long short-term memory network model; adjusting the cooling strategy and discharge current according to the load prediction value output by the model; transmitting the data to the cloud platform to optimize the model parameters; updating the operation parameters of the device according to the feedback of the cloud platform.

[0042] Embodiment 1: This embodiment provides a management method for a forklift power battery pack based on dynamic load prediction and multi-scenario optimization. Taking a large-scale logistics and warehousing environment as the application scenario, the specific implementation manner of the technical solution is demonstrated. For example, in a certain logistics park, forklifts need to perform frequent loading and unloading operations, and the operation scenarios include: High-frequency start-stop operation scenario: used to transport goods from the warehouse shelves to the loading and unloading area, and the frequent start-stop during the operation causes significant current fluctuations in the battery pack; Heavy-load long-distance transportation scenario: used to carry heavy goods to the remote storage area, and the battery pack needs to bear high load for a long time, resulting in rapid battery temperature rise; Low-temperature environment operation scenario: outdoor goods loading and unloading operations under low-temperature conditions in winter, and the battery performance is significantly affected by low temperature.

[0043] To solve the problems in the above scenarios, this embodiment adopts the following steps: By collecting forklift operation data in real time, including current, voltage, load weight and position information, load prediction is carried out based on the bidirectional LSTM (Long Short-Term Memory Network) model. The LSTM model inputs historical working condition data and current real-time data to predict the load demand in the short term in the future. For example, in the frequent start-stop scenario, the prediction module anticipates in advance the current impact that the battery pack may suffer, and combines the predicted value optimized by the load weight coefficient to guide the adaptive current control module to dynamically adjust the discharge path, balance the load of battery cells, and avoid current overshoot. In the heavy-load long-distance transportation scenario, the system dynamically selects a suitable cooling mode (air cooling or liquid cooling) based on the future load demand value provided by the load prediction module. After experimentally calibrating the cooling coefficient, combining the real-time temperature and the load prediction value, the cooling power is adjusted to optimize the cooling energy consumption and temperature rise control; specifically, at the initial stage of high-load operation, the system preferentially adopts the efficient liquid cooling mode to quickly reduce the battery temperature rise; when the load decreases, the system switches to the low-energy air cooling mode to achieve the dynamic balance of cooling efficiency and energy consumption.

[0044] In a low-temperature environment, the system combines load prediction and battery internal impedance calculation. By optimizing the discharge current path, it preferentially selects battery cells with lower impedance to participate in discharge, reducing the impact of battery temperature rise on performance. At the same time, by adjusting the current distribution in real-time feedback, it ensures the discharge uniformity of all battery cells and extends the battery life. And the system transmits the operation data to the cloud platform through the Internet of Things module. The cloud platform conducts long-term analysis on the load prediction error, cooling energy consumption, and battery life data during operation, and iteratively optimizes the weights of the load prediction model and the thermal management parameters. For example, in a frequent start-stop scenario, the cloud platform adjusts the model parameters according to the historical data of high load impacts to optimize the prediction accuracy. In a low-temperature scenario, it optimizes the cooling coefficient through the historical temperature rise curve to improve the heat dissipation effect. Through this embodiment, the following significant improvements have been achieved in the actual application in the logistics park: In high-frequency start-stop operations, the current impact is reduced by about 20%, and the battery cell life is extended by 15%. In heavy-load scenarios, the dynamic switching between the liquid cooling mode and the air cooling mode reduces the cooling energy consumption by 25%, and the battery temperature rise is reduced by about 30%. In a low-temperature environment, the discharge efficiency is increased by 10%, and the operation stability is significantly enhanced. The optimization and iteration of the cloud platform enable the overall system to have the ability of continuous learning and improvement in complex scenarios, showing higher intelligent adaptability.

[0045] Embodiment 2: This embodiment optimizes the cooling mode switching conditions in the multi-mode thermal management module to enhance its applicability in extreme high-temperature and low-temperature working conditions. Specifically, in the cooling mode selection logic, based on the battery pack temperature and the load prediction value , the cooling mode is dynamically selected: When , the passive heat dissipation mode is maintained; when , the air cooling mode is activated, and the cooling power adjustment formula is: ; when , the air cooling mode is preferentially used, but liquid cooling is dynamically introduced as a supplement, and the mixed cooling power is calculated as: ; where represents the air cooling ratio, which is dynamically adjusted according to the real-time load prediction value, represents the cooling power in the air cooling mode, represents the cooling power in the liquid cooling mode; when , the liquid cooling mode is fully activated, and the cooling intensity is dynamically adjusted according to the load prediction value: ; the calibration method of its cooling coefficient is that the calibration of the cooling coefficient and is completed through the following steps: Simulate different temperature and load conditions in the laboratory environment, record the heat dissipation efficiency and energy consumption data; optimize the cooling coefficient according to the ratio of the minimum energy consumption and the best heat dissipation to ensure that the model is applicable to different working conditions.

[0046] Meanwhile, to eliminate ambiguity and further clarify the source and meaning of the parameters in the formula, for example, in the optimization of the load prediction formula, the weight coefficients in the prediction model and are determined by the following method: trained through the gradient descent algorithm, and the initial weights are optimized using historical load data; the weight update is based on the following formula: ; where is the learning rate, is the prediction error function, represents the weight at the current moment, represents the updated weight, is the model weight matrix. Also, the temperature control threshold is optimized by simulating the temperature rise curve in the actual working conditions to ensure the applicability of the temperature control strategy in high-frequency start-stop, heavy-load, and low-temperature scenarios. The edge computing module further enhances the data collaboration ability in this embodiment, such as real-time data processing. A distributed computing framework is introduced on the edge device to achieve the following functions: dynamically aggregating sensor data (including voltage, current, temperature, and position information); accelerating the execution of the short-term prediction model using the local cache mechanism; and in terms of collaborative optimization between modules, such as the interaction between the dynamic load prediction module and the multi-mode thermal management module is managed by a unified data flow protocol to reduce latency; the edge computing device periodically transmits the optimization parameters (such as the cooling coefficient and the prediction model weight) to the cloud platform for long-term optimization; an adaptive fault handling mechanism for the edge computing module is added. When a certain module fails to work properly, the system automatically switches to the backup strategy. The temperature control strategy in this embodiment is more flexible: covering a wider temperature range, significantly reducing the battery temperature rise and cooling energy consumption; the edge computing ability is improved: the data processing efficiency and the collaborative optimization effect between modules are improved, and the system stability is enhanced.

[0047] Embodiment 3: This embodiment specifically illustrates the source and calibration method of the variable parameters, as well as the specific implementation path of the cloud platform optimization model. In terms of multi-dimensional cooling coefficient calibration, for example, three typical working conditions of high-frequency start-stop, heavy-load long-distance transportation, and low-temperature environment are simulated in the laboratory, and the temperature rise curve and energy consumption data under each working condition are recorded by the edge computing device. The cooling coefficient and are calculated using the following formula: ; where represents the energy consumed during the cooling process, is the temperature change, is the predicted value of the load; according to the experimental results, optimize the cooling coefficient to make the heat dissipation efficiency and energy consumption ratio reach the optimum. In terms of the implementation of the cloud platform optimization model, such as data upload and analysis: the operation data collected in real time by the edge computing module, including the load prediction error, the cooling power usage, and the battery life data, are periodically transmitted to the cloud platform; weight update mechanism: based on the gradient descent algorithm, the cloud platform optimizes the model weights, and the formula is as follows: , where is the learning rate, is the load prediction error function, is the model weight, represents the model weight before optimization, represents the model weight after optimization. Cooling strategy adjustment: combine the historical operation data to optimize the temperature control threshold for cooling mode switching. For example, adjust the threshold in the low-temperature scenario: ; ensure that the cooling power adapts to the low-temperature environment requirements. In terms of the parameter range and value, such as the cooling coefficient The value range of: 0.1 - 2.0, which is applicable to the heat dissipation requirements under different load conditions; the learning rate for weight update: the value range is 0.001 - 0.01, ensuring the stable convergence of model optimization; the operation data collection period: 1 second, ensuring real-time and data integrity. Also, in terms of edge computing and module collaboration, such as real-time fault handling: when the edge computing device detects a sensor anomaly, it automatically switches to the redundant module to avoid data loss; data flow closed-loop optimization: the prediction module and the thermal management module interact through a unified protocol to reduce data latency. To verify the technical effects of this embodiment, the experimental scenarios of the present invention include three typical working conditions: high-frequency start-stop, heavy-load transportation, and low-temperature operation, and the detailed settings are as follows: High-frequency start-stop scenario: The forklift performs repeated cargo loading and unloading within 300 meters, simulating the current impact caused by high-frequency start-stop. The collected data includes the battery unit voltage, current change rate, and temperature rise curve; Heavy-load transportation scenario: The forklift transports heavy goods exceeding 50% of the rated load and travels a distance of more than 1 kilometer to evaluate the adaptability of load prediction and cooling management; Low-temperature operation scenario: Outdoor operation is carried out under the condition of an ambient temperature of -10°C to test the efficiency of the thermal management system and the stability of battery performance.

[0048] It can be obtained from the experimental results that for cooling energy consumption optimization: in the heavy-load transportation scenario, based on the cooling coefficient , Dynamic adjustment enables the system to successfully achieve efficient switching between air cooling and liquid cooling. The cooling energy consumption is reduced by 20% compared to the existing technology, and the temperature rise is controlled within 10°C, significantly reducing the risk of battery overheating. Battery life extension: The adaptive current control module equalizes the discharge of battery cells. Experimental data shows that the battery life is extended by 15% in high-frequency start-stop scenarios, reducing cell loss caused by current overshoot. Load prediction accuracy: The dynamic load prediction module combines the long short-term memory (LSTM) algorithm. Based on 3000 sets of load data collected from experiments, the load prediction accuracy is improved from 87% of traditional methods to 95%, significantly reducing the impact of prediction deviation on the cooling strategy. Optimization efficiency improvement: The collaboration optimization cycle between the edge computing device and the cloud platform is shortened from 72 hours to 48 hours, mainly due to the real-time tuning of cooling parameters by the edge computing module and the rapid iteration ability of the cloud platform. And during the experiment, all data is collected by the edge computing device and transmitted to the cloud platform for analysis. The cooling energy consumption is recorded by the real-time power output of the thermal management module, the battery life is estimated from the change trends of cell voltage and internal impedance, and the load prediction accuracy is calculated based on the error between the predicted value and the actual load.

[0049] Example 4: In this example, the dynamic load prediction module adopts a dynamic load prediction model based on the long short-term memory (LSTM) algorithm, and further refines the parameter optimization process and data preprocessing method of the model. To improve the accuracy of load prediction, the following key technical steps are introduced: Historical data normalization: All input load data (current, voltage, temperature, forklift operation load) is first normalized to eliminate the dimensional differences between different data sources and ensure the efficient training of the LSTM model. Data segmentation processing: Real-time load data is segmented according to time series, and the load fluctuation frequency and amplitude within each time period are calculated respectively as the input of the LSTM model. By analyzing the load data in multiple time periods, the complexity of single calculation can be reduced, and the calculation speed and real-time performance can be improved. LSTM model parameter optimization: The gradient descent algorithm is used to optimize the model weights and update the historical load weights and real-time data weights in the LSTM model. The weight update formula is as follows: ; where is the learning rate, is the load prediction error function. This process dynamically adjusts the weights in the LSTM model through the backpropagation algorithm to optimize the prediction accuracy. Load prediction value calculation: According to the output of the LSTM model, the load demand for future time periods is generated, and the prediction result is dynamically adjusted through the following formula: ; where , is the hidden state at the previous moment, is the input data at the current moment, is an activation function, and the definitions and functions of each variable have been clearly described above.

[0050] In addition, to better adapt to complex working environments, the present embodiment has refined the cooling mode selection and adjustment of the multi-mode thermal management module to ensure the optimal balance between energy efficiency and thermal control under different load conditions. The basis for cooling mode selection: The selection of the cooling mode is based on the real-time load prediction value and the current temperature of the battery pack , and the specific cooling mode selection logic is as follows: When , the passive cooling mode is enabled. When , the air cooling mode is enabled, and the cooling power is adjusted according to the formula: ; where is the cooling coefficient of the air cooling mode. When , the liquid cooling mode is enabled, and the cooling intensity is dynamically adjusted according to the load prediction value: ; where is the load-related cooling coefficient. Calibration of the cooling coefficient: The cooling coefficients and are obtained through experimental calibration. In the experiment, by simulating different temperature and load conditions, the heat dissipation efficiency and energy consumption in each cooling mode are recorded, and the cooling coefficients are optimized to achieve the best balance between the heat dissipation effect and energy consumption. And in order to further improve the performance of the battery pack and extend the battery life, the present embodiment has optimized the adaptive current control module. The specific implementation steps are as follows: Discharge current adjustment: According to the real-time load prediction value and the state parameters of the battery cells, the discharge current of each battery cell is adjusted in real time to balance the load distribution among the battery cells. The specific adjustment formula is: ; where is the number of battery cells,

[0051] Battery internal resistance compensation: Considering the influence of the change in the internal impedance of the battery on the discharge current, the present embodiment further introduces an internal resistance compensation mechanism. Specifically, the internal resistance of the battery cell is calculated by real-time monitoring of the battery voltage and the current , and the change in the internal resistance will cause the correction of the discharge current. The internal resistance calculation formula is: ; where is the load terminal voltage. This mechanism can effectively avoid the influence of battery internal resistance change on current control and maintain the stability of the discharge process. In this embodiment, further optimization is made in the collaborative optimization between modules to ensure the high efficiency and no delay of the cooperation of each module. The specific optimization steps are as follows: Data transfer mechanism between modules: Each module (dynamic load prediction, thermal management, current control) works collaboratively through a unified data flow protocol. The load prediction module updates the prediction result every 1 second and transfers it to the thermal management module for real-time adjustment of the cooling strategy. The thermal management module adjusts the cooling mode in real time according to the load prediction result, and the cooling efficiency is fed back to the adaptive current control module to dynamically optimize the discharge current of the battery unit. Optimization of edge computing: The edge computing module is responsible for real-time summarizing, processing, and transferring all sensor data to ensure low latency and high efficiency of data processing. The collaborative optimization between modules is accelerated through the local computing function of the edge device, reducing the data transmission time and ensuring the real-time response of the system.

[0052] The introduction of the cloud platform enables the system to perform real-time optimization according to the data during operation. The specific operation steps are as follows: Data upload and analysis: The data (including load prediction error, cooling efficiency, battery life data, etc.) collected in real time through the edge computing module is periodically transferred to the cloud platform. The cloud platform analyzes these data and continuously optimizes the load prediction model and thermal management strategy. Optimization mechanism: Based on the historical data analyzed by the cloud platform, the gradient descent algorithm is used to dynamically adjust the weights of the load prediction model and the cooling coefficients of the thermal management strategy. The model update formula is as follows: ; Through continuous learning and optimization, the cloud platform can improve the prediction accuracy and energy efficiency of the system, enabling the system to always maintain the best performance under complex working conditions.

[0053] Embodiment 5: This embodiment further refines the specific implementation path of the collaborative optimization of dynamic load prediction and multi-mode thermal management in the forklift power battery pack management method, especially in the calibration of cooling coefficients, data collaboration between modules, and optimization of the load prediction model, ensuring that the functions and roles of each module are fully demonstrated and strengthening the operability and realizability of the system. To improve the cooling efficiency and ensure the stability of the battery pack under different load and temperature conditions, this embodiment dynamically calibrates the cooling coefficients. The cooling coefficients and respectively affect the cooling effect and energy consumption in the air-cooled and liquid-cooled modes. The present invention completes the calibration of the cooling coefficients through the following steps: Simulate different temperature and load conditions in the laboratory and record the heat loss and energy consumption data in each cooling mode. Specifically, under different working conditions of temperature and load prediction value , determine the optimal cooling coefficients and , and its calculation formula is as follows: ; where is the energy consumed during the cooling process, is the temperature change, is the load prediction value. This method ensures the balance between cooling effect and energy consumption. In practical applications, the system adjusts the cooling intensity in real time according to the temperature and the load prediction value . For example, when , the system activates the liquid cooling mode and adjusts the cooling intensity according to the load prediction value: ; This adjustment mechanism ensures that the battery pack is cooled in a timely manner under high load or extreme temperature conditions, avoiding overheating or energy waste. In this embodiment, the load prediction module uses the Long Short-Term Memory (LSTM) algorithm for prediction. To improve the prediction accuracy and system adaptability, the present invention proposes the following optimization steps: normalize the historical load data to ensure the dimensional consistency between different data sources. Train the load prediction model using historical data and update it dynamically in combination with real-time data. The weight coefficients and are optimized and updated by the gradient descent method, and the update formula is: ; where is the learning rate, is the load prediction error function. This formula ensures the real-time adjustment ability of the model under different working conditions. The load prediction value will directly affect the selection and adjustment of the cooling strategy. By combining real-time temperature and load prediction, the system can dynamically select the air cooling or liquid cooling mode and adjust the cooling coefficient and cooling intensity, thereby optimizing the operating efficiency of the battery pack.

[0054] The collaborative optimization between modules is one of the key innovations of the present invention. The dynamic load prediction module and the multi-mode thermal management module work together to achieve delay-free data transmission through a unified data stream protocol. The specific steps are as follows: update the load prediction value once per second and transmit it to the multi-mode thermal management module. The multi-mode thermal management module adjusts the cooling strategy according to the load prediction value and feeds back the actual cooling efficiency to the adaptive current control module. The adaptive current control module adjusts the discharge current of the battery unit according to the feedback cooling efficiency to ensure the stability of the battery under different loads and temperatures. To reduce system latency and improve data processing efficiency, this embodiment uses edge computing technology to perform real-time processing and collaborative optimization of the data of each module. The edge computing module is responsible for receiving and processing data from sensors, including information such as voltage, current, and temperature, and accelerating the calculation of the load prediction model through local caching. All optimization results will be periodically transmitted to the cloud platform for long-term data recording and model update.

[0055] The system described in this embodiment includes the following hardware components. Sensor module: used to collect the voltage, current, temperature of the battery pack and the forklift operation load data in real time. Edge computing module: used for local data processing, supporting real-time load prediction and cooling management. Cloud platform: used for long-term data storage and model optimization, supporting the continuous improvement of system performance. In the specific steps of its implementation path, data collection: the status data of the battery pack is collected in real time through sensors and transmitted to the edge computing module. Load prediction and cooling management: according to the load prediction results and real-time temperature, the edge computing module transmits the data to the thermal management module for cooling strategy adjustment. Current control: according to the load prediction and temperature control feedback, the current control module adjusts the discharge current of the battery unit in real time to maintain the balanced discharge of the battery. System feedback and optimization: all operation data is regularly transmitted to the cloud platform for historical data analysis to optimize the prediction model and cooling strategy.

[0056] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A forklift power battery pack management method, characterized in that: The method comprises the following management steps: Step 1, real-time data collection through a data collection module: real-time data collection of voltage, current, temperature of a battery pack and operation load data of a forklift through a distributed sensor module, wherein: voltage data is used to evaluate the state of a battery cell; current data is used to calculate the instantaneous load demand of a forklift; temperature data is used to monitor the thermal stability of the battery pack; and operation load data is dynamically recorded based on the operation mode of the forklift; Step 2: Perform dynamic load forecasting through the dynamic load forecasting module: Based on the long short-term memory network (LSTM) algorithm, establish a forecasting model to segment the real-time collected load data; use historical load data to train the model weights, and update the model parameters in real time to optimize the forecasting accuracy; predict the load demand for the next period, specifically quantify the load fluctuation amplitude and frequency, and generate a load forecast value , the calculation formula is: ;in, and They are historical data weight and real-time data weight. is the hidden state value at the previous moment, is the current input data, is the bias term, is the activation function; Step 3: Perform multi-mode thermal management through the multi-mode thermal management module: Dynamically select the cooling mode based on real-time temperature data and load prediction values: When the temperature When the temperature is less than 30°C, the passive cooling mode is activated; when the temperature When the temperature is between 30°C and 45°C, the air cooling mode is activated and the cooling power is adjusted according to the formula: ;in, is the cooling power adjustment coefficient; when the temperature When the temperature exceeds 45°C, liquid cooling mode is activated and the cooling intensity is dynamically adjusted according to the load prediction value : ;in, is the load-dependent cooling factor; Step 4: Adaptive current control is performed through the adaptive current control module: according to the load prediction value and battery pack status parameters to adjust the discharge current of each battery cell in real time , to balance the load distribution among battery cells, the specific control rules are: in, is the number of battery cells, is the unit battery voltage; limits the unit battery current change rate to ensure that the change within one second does not exceed 5% of the rated value; Step 5, inter-module collaborative optimization: The dynamic load prediction module works with the multi-mode thermal management module to adjust the cooling mode in real time according to the load prediction value; the adaptive current control module dynamically optimizes the battery cell discharge strategy based on the load prediction and thermal management feedback; data is transmitted through a unified edge computing device; Step 6, data feedback and system update: Data is transmitted to the cloud platform through edge computing devices. The cloud platform analyzes and predicts model performance and thermal management effects, and regularly updates model weights and cooling strategy parameters. The cloud platform generates a dynamic maintenance plan, including battery maintenance recommendations and cooling system optimization guidance. The weight coefficients of the prediction model and The initial value of is determined by the following method: normalizing the historical load data; fitting the prediction error using the least squares method and calculating the gradient of each iteration; optimizing the weight coefficient based on the gradient descent algorithm until the prediction error converges to the preset threshold; Cooling power adjustment coefficient of the multi-mode thermal management module and load-dependent cooling factor It is obtained through experimental calibration methods, including: measuring the heat loss at the cooling mode switching point under different temperature conditions; calculating the energy consumption and heat dissipation ratio under each cooling mode, and optimizing the cooling coefficient; The discharge current of the adaptive current control module The adjustment strategy further includes: correcting the discharge current in real time according to the internal impedance change of the battery cell; the internal impedance change is calculated by collecting the voltage and current of the battery cell in real time, and the formula is: ;in, is the load terminal voltage.

2. The forklift power battery pack management method according to claim 1, characterized in that: The long short-term memory network LSTM model in the dynamic load forecasting further includes: an embedding layer for reducing the dimension of the load data collected in real time; a bidirectional LSTM layer for simultaneously capturing the before and after time dependencies of the load data; and a dense connection layer for converting the output time series forecast value into the load forecast value at the next moment.

3. The forklift power battery pack management method according to claim 1, characterized in that: The collaborative optimization between the modules further includes the following steps: the dynamic load prediction module updates the prediction value every 1 second and transmits the prediction result to the multi-mode thermal management system; the multi-mode thermal management system updates the cooling strategy according to the load prediction value, and feeds back the actual cooling efficiency to the adaptive current control module; the adaptive current control module adjusts the discharge current in real time, and returns the adjustment result to the dynamic load prediction module to update the load model at the next moment.

4. The forklift power battery pack management method according to claim 1, characterized in that: During the data feedback and system update, the following data are recorded through the cloud platform for model optimization: the actual error of each load prediction; the energy consumption and heat dissipation efficiency under each cooling mode; the discharge curve and life trend of the battery unit.

5. A device based on the forklift power battery pack management method according to claim 1, characterized in that: include: Data acquisition module, used to collect voltage, current, temperature of battery pack and forklift operation load data in real time; The prediction module is used to predict the load demand at the next moment based on the long short-term memory network (LSTM) model; the thermal management module is used to dynamically adjust the cooling mode according to the load prediction value and temperature data; the current control module is used to adjust the discharge current of the battery unit in real time; the edge computing module is used to realize data transmission and collaborative optimization between modules.

6. A device based on the forklift power battery pack management method according to claim 1, characterized in that: The device includes: at least one device, which includes a data acquisition module, a prediction module, a thermal management module, a current control module and an edge computing module; a communication module, which is used to communicate with a cloud platform to upload data and send feedback; and a storage module, which is used to store load data, temperature data and model optimization parameters.

7. A storage medium based on the forklift power battery pack management method according to claim 1, characterized in that: The storage medium stores computer-readable instructions. When the instructions are executed by a computing device, the computing device implements the following steps: collecting real-time data of the battery pack and inputting it into a long short-term memory network model; adjusting the cooling strategy and discharge current according to the load prediction value output by the model; transmitting the data to a cloud platform to optimize the model parameters; and updating the operating parameters of the device according to feedback from the cloud platform.

Images