Equalization charging method, device and equipment for power battery pack

Abstract

The invention relates to the technical field of battery management, and discloses an equalizing charging method, device and equipment for a power battery pack. The method comprises the following steps: performing parameter acquisition on a power battery pack, measuring terminal voltage values, temperature values and charge and discharge current values of each single battery, and calculating a state-of-charge value and an internal resistance change rate of each single battery; inputting into a multi-head self-attention network for feature extraction, and generating a state correlation feature matrix; constructing an Actor network and a Critic network based on the state correlation characteristic matrix, and outputting an equalization current value, a constant-current or constant-voltage equalization mode and equalization time through the Actor network; according to the equalization current value, the equalization mode and the equalization time, the DC-DC converter in the main equalization loop is controlled to generate constant current output, or the Buck-Boost converter in the secondary equalization loop is controlled to generate constant voltage output, the equalization efficiency of the power battery pack is guaranteed, meanwhile, energy loss is reduced, and temperature self-adaptive control over the equalization circuit is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery management, and in particular to a balanced charging method, device and equipment for a power battery pack. Background Art

[0002] With the rapid development of the new energy vehicle industry, the performance and life of power battery packs as core components have attracted much attention. In actual applications, due to the influence of factors such as manufacturing process, ambient temperature, and operating conditions, the performance parameters of each single cell in the power battery pack will gradually differ. This inconsistency will cause some cells to be overcharged or over-discharged, seriously affecting the service life and safety of the battery pack.

[0003] Traditional equalization charging methods mainly use passive equalization or simple active equalization strategies. Although these methods can improve the consistency of cells to a certain extent, the equalization efficiency is low and cannot adapt to the dynamic change characteristics of the battery pack under different temperatures, charge and discharge rates and cycle times. Especially for large-scale parallel power battery packs, due to the large number of cells, traditional equalization methods are difficult to accurately identify the cells that need to be balanced, which easily leads to waste of equalization resources. Summary of the invention

[0004] The present invention provides a balanced charging method, device and equipment for a power battery pack, which reduces energy loss while ensuring the balanced efficiency of the power battery pack and realizes temperature adaptive control of a balanced circuit.

[0005] In a first aspect, the present invention provides a balanced charging method for a power battery pack, the balanced charging method for a power battery pack comprising: Collect parameters of the power battery pack, measure the terminal voltage, temperature, and charge and discharge current of each single battery, and calculate the state of charge and internal resistance change rate of each single battery; The terminal voltage value, the temperature value, the charge and discharge current value, the state of charge value and the internal resistance change rate are constructed into a state feature sequence, and input into a multi-head self-attention network for feature extraction to generate a state correlation feature matrix; Based on the state association feature matrix, a three-layer fully connected Actor network and a dual-Q structure Critic network are constructed, and a balanced current value, a constant current or constant voltage balanced mode and a balanced time are output through the Actor network; According to the balancing current value, the balancing mode and the balancing time, the DC-DC converter in the main balancing loop is controlled to generate a constant current output, or the Buck-Boost converter in the secondary balancing loop is controlled to generate a constant voltage output.

[0006] In a second aspect, the present invention provides a balanced charging device for a power battery pack, the balanced charging device for the power battery pack comprising: The parameter acquisition module is used to collect parameters of the power battery pack, measure the terminal voltage value, temperature value and charge and discharge current value of each single battery, and calculate the state of charge value and internal resistance change rate of each single battery; A feature extraction module, used to construct the terminal voltage value, the temperature value, the charge and discharge current value, the state of charge value and the internal resistance change rate into a state feature sequence, and input it into a multi-head self-attention network for feature extraction to generate a state correlation feature matrix; A construction module, used to construct a three-layer fully connected Actor network and a Critic network with a dual-Q structure based on the state association feature matrix, and output a balanced current value, a constant current or constant voltage balanced mode and a balanced time through the Actor network; The control module is used to control the DC-DC converter in the main balancing loop to generate a constant current output, or control the Buck-Boost converter in the secondary balancing loop to generate a constant voltage output according to the balancing current value, the balancing mode and the balancing time.

[0007] A third aspect of the present invention provides a balanced charging device for a power battery pack, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the balanced charging device for the power battery pack executes the above-mentioned balanced charging method for the power battery pack.

[0008] In the technical solution provided by the present invention, by introducing a multi-head self-attention network for feature extraction, efficient processing of power battery pack state information is achieved, the correlation between different monomers is accurately captured, and the expression ability of state features is improved. The Actor-Critic deep reinforcement learning framework is adopted to dynamically generate a combination strategy of balancing current, balancing mode and balancing time, and adaptive adjustment of balancing control is achieved, so that the balancing strategy can be optimized according to the real-time state of the battery pack. A multi-level balancing circuit structure in which the main balancing loop and the secondary balancing loop work together is designed, and a balancing mode switching mechanism based on voltage difference is established, which reduces energy loss while ensuring balancing efficiency. A complete balancing charging protection system is constructed, and the safety and reliability of the balancing charging process are ensured by real-time monitoring of monomer voltage, temperature and balancing current, and setting multi-level protection thresholds. The proximal strategy optimization algorithm is adopted to update the network parameters online, and the exploration efficiency of reinforcement learning is improved by introducing advantage functions and strategy entropy, and the convergence performance of the algorithm is enhanced. The temperature adaptive control of the balancing circuit is realized, and the overheating of the device is effectively prevented through real-time monitoring of the junction temperature of the switch tube and dynamic adjustment of the duty cycle, which prolongs the service life of the system. Through the comprehensive evaluation of voltage standard deviation, temperature distribution extreme difference and charge state consistency indicators, quantitative evaluation and strategy optimization of balancing effect are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0010] Figure 1 A schematic diagram of a flow chart of a balanced charging method for a power battery pack provided in an embodiment of the present application; Figure 2 A schematic block diagram of the structure of a balanced charging device for a power battery pack provided in an embodiment of the present application; Figure 3 A schematic block diagram of the structure of a balanced charging device for a power battery pack provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change based on actual conditions.

[0013] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0014] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0015] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0016] See also Figure 1 , Figure 1 A schematic diagram of a flow chart of a balanced charging method for a power battery pack provided in an embodiment of the present application, such as Figure 1 As shown, the balanced charging method of the power battery pack provided in the embodiment of the present application includes steps S100 to S600.

[0017] Step S100, collecting parameters of the power battery pack, measuring the terminal voltage value, temperature value and charge and discharge current value of each single battery, and calculating the state of charge value and internal resistance change rate of each single battery; It is understandable that the execution subject of the present invention may be a balanced charging device for a power battery pack, or a terminal or a server, which is not specifically limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0018] Specifically, the sampling timing is set for multiple sampling points in the power battery pack to obtain a parameter acquisition timing matrix, which is used to coordinate the data acquisition of multiple sensors so that the data at different time points can be compared and analyzed on the same time axis. Since the power battery pack is composed of multiple cells connected in series and parallel, the state of each cell is different. It is necessary to ensure that the construction of the sampling timing matrix meets the requirements of balanced charging and avoid the failure of the control algorithm due to data asynchrony. During the sampling process, the voltage data is collected at the sampling points according to the parameter acquisition timing matrix, and the terminal voltage value of each single cell is measured using a high-precision voltage sensor, and the ADC (analog-to-digital converter) is used for digital processing for subsequent feature extraction and calculation. At the same time, the temperature data is collected at the sampling points according to the parameter acquisition timing matrix, and the temperature value of the battery under the working state is collected using a thermistor or temperature sensor, and the temperature is corrected in combination with the battery thermal management system to ensure the accuracy of the temperature data. At the same time, the current data is collected at the sampling points according to the same parameter acquisition timing matrix, and the charging and discharging current values ​​are measured using a Hall current sensor or a shunt resistor, and the collected original signal is filtered to eliminate the influence of high-frequency noise on the measurement results. After obtaining the charge and discharge current value, a time integration operation is performed to calculate the state of charge (SOC) of the battery. The state of charge is calculated using the Coulomb integration method, that is, by integrating the charge and discharge current over time, the accumulated charge is obtained, and it is superimposed with the initial state of charge value to calculate the remaining charge of the battery. The open circuit voltage characteristics of the battery are combined to make corrections to improve the accuracy of the SOC calculation and obtain the target value. The target value is standardized and normalized with the nominal capacity of the battery to obtain a relative state of charge value between 0 and 1. In the process of calculating the internal resistance change rate, the terminal voltage value and the charge and discharge current value are differentially calculated at adjacent sampling points to obtain the voltage change rate and current change rate of the battery in a short time. This differential operation can reflect the dynamic characteristics of the battery, especially in pulse discharge or high current charge and discharge, and can effectively extract the instantaneous impedance information of the battery. The differential result is processed by Kalman filtering to reduce the measurement error and enhance the signal stability. Kalman filtering dynamically estimates the measured value by recursive updating, thereby improving the calculation accuracy of the battery internal resistance change rate. After filtering, the differential signal is ratio-calculated, and the internal resistance change rate of the battery is estimated by calculating the ratio of the voltage change rate to the current change rate, reflecting the dynamic evolution of the internal impedance of the battery.

[0019] Step S200, constructing the terminal voltage value, temperature value, charge and discharge current value, state of charge value and internal resistance change rate into a state feature sequence, and inputting it into a multi-head self-attention network for feature extraction to generate a state correlation feature matrix; Specifically, the terminal voltage value, temperature value, charge and discharge current value, state of charge value and internal resistance change rate are feature spliced ​​and dimensionally transformed to obtain a state feature sequence. Before splicing, each feature is normalized to ensure that the scales of different physical quantities match and to avoid excessive influence of specific features on model learning. After normalization, all features are spliced ​​in time series to form a state feature sequence, which reflects the overall state evolution of the power battery pack. The state feature sequence is linearly transformed to generate a query matrix (Query), a key matrix (Key) and a value matrix (Value). The original features are projected into different feature spaces using a learnable weight matrix to enhance the model's ability to understand the relationship between different features. By linearly transforming the state feature sequence, the query matrix, the key matrix and the value matrix are obtained respectively, and matrix multiplication is performed on this basis, and the query matrix and the key matrix are dot-producted to obtain the attention score matrix. The attention score matrix is ​​scaled, that is, divided by the square root of the matrix column dimension to prevent the occurrence of gradient vanishing or gradient explosion. After completing the matrix multiplication, the attention score matrix is ​​subjected to Softmax normalization to ensure that the sum of weights between different sampling points is 1, and the attention weight matrix is ​​obtained. The attention score is converted into a probability distribution, so that the model learns the importance of each feature at different time points and dynamically adjusts the weight distribution. The attention weight matrix is ​​matrix multiplied with the value matrix to calculate the output features of the single-head attention and extract the temporal dependency between different state parameters of the power battery pack. The state feature sequence is input into multiple parallel attention calculation units in the multi-head self-attention network respectively, and each attention calculation unit independently performs attention feature extraction to obtain multiple different attention output features. Different attention calculation units use independent query, key and value projection matrices, and each head pays attention to different feature relationships, so that the model learns the state evolution pattern of the power battery pack from multiple angles. For each attention calculation unit, the calculation process is similar to the single-head attention mechanism, including linear transformation, matrix multiplication, scaling, Softmax normalization and final weighted summation. Since N attention calculation units independently calculate N different attention output features, the N attention output features are spliced ​​in the final stage to construct a richer feature vector. After the concatenation operation is completed, the target feature vector is normalized by the LayerNorm layer to improve training stability and accelerate model convergence. The role of LayerNorm is to normalize the features within each sample so that the model maintains a good learning effect under different batches of data input. The normalized target feature vector eventually forms a state-related feature matrix, which contains state information such as the terminal voltage, temperature, current, state of charge, and internal resistance change rate of each single cell in the power battery pack, and extracts the correlation between these features through a multi-head self-attention mechanism.

[0020] Step S300: construct a three-layer fully connected Actor network and a double-Q structured Critic network based on the state association feature matrix, and output a balanced current value, a constant current or constant voltage balanced mode, and a balanced time through the Actor network; Specifically, the state-related feature matrix is ​​input into the first fully connected layer of the Actor network. The first fully connected network contains 512 neurons, and the ReLU activation function is used for nonlinear transformation to obtain the first layer of action features. The role of the ReLU function is to introduce nonlinearity so that the model can learn complex state mapping relationships. Since ReLU has good gradient propagation characteristics on the positive semi-axis, it can avoid the gradient vanishing problem and ensure the effective training of deep networks. The first layer of action features is input into the second fully connected layer of the Actor network, which contains 256 neurons and is also nonlinearly transformed using the ReLU activation function. This layer further extracts more abstract action features, so that the network can more accurately fit the complex mapping relationship of the equilibrium strategy and obtain the second layer of action features. The second layer of action features is input into the third fully connected layer of the Actor network, which contains 32 neurons and is nonlinearly transformed using the Tanh activation function to generate action output features. The Tanh activation function limits the output value to the range of [−1,1], ensuring that the output of the Actor network has a reasonable numerical range, and since the Tanh function has a large gradient near the zero point, it helps to improve learning efficiency. The action output features are input into three parallel action mapping layers to calculate the balancing current value, balancing mode and balancing time respectively. Among them, the output of the first action mapping layer corresponds to the balancing current value, which is used to control the current regulation of the power battery pack during the balancing charging process to ensure that the voltages of different battery cells tend to be consistent. The output of the second action mapping layer is used to determine the balancing mode, that is, to choose between the constant current mode and the constant voltage mode, where the constant current mode is suitable for battery balancing with a large voltage difference, while the constant voltage mode is more suitable for refined balancing control to reduce battery loss and improve charging efficiency. The output of the third action mapping layer determines the balancing time, that is, to control the duration of the balancing current or voltage action to achieve accurate balancing and avoid over-compensation or under-compensation. At the same time, after the Actor network generates the action output features, the value of the action is evaluated, and the state-related feature matrix and the action output features are combined and input into two Q networks with the same structure, that is, the dual Q structure of the Critic network. Each Q network contains three fully connected layers, the first fully connected layer contains 512 neurons, the second fully connected layer contains 256 neurons, and the third fully connected layer contains only one neuron, and finally outputs a Q value. All layers use ReLU activation function for nonlinear transformation to ensure that the Critic network effectively fits the value function of the state-action pair. During the calculation process, each Q network independently evaluates the input state and action, and calculates the first Q value and the second Q value respectively. The introduction of the double Q structure is to reduce the overestimation bias problem in the strategy update process, so that the Actor network can learn a more stable and reliable equilibrium strategy during the training process.The first Q value and the second Q value are selected for minimum value, that is, the smaller value is selected from the two Q values ​​as the target Q value, thereby reducing the over-optimism problem of Q value estimation and improving the stability of the balancing strategy. The target Q value is used as a value evaluation indicator for the output action of the Actor network to guide the training of the Actor network so that it can continuously optimize the balancing strategy and ensure that the selection of balancing current, balancing mode and balancing time can maximize the overall performance of the power battery pack.

[0021] Step S400: According to the balancing current value, balancing mode and balancing time, the DC-DC converter in the main balancing loop is controlled to generate a constant current output, or the Buck-Boost converter in the secondary balancing loop is controlled to generate a constant voltage output.

[0022] Specifically, the balancing current value and the voltage difference between adjacent cells are compared and analyzed to determine the selection of the balancing circuit. The terminal voltage value of each cell is obtained, and the voltage difference between adjacent cells is calculated, and then compared with the preset voltage value. If the voltage difference is greater than the preset voltage value, it means that there is a large voltage imbalance. At this time, the main balancing circuit is selected for balancing charging to quickly reduce the voltage deviation between the battery cells; if the voltage difference is less than or equal to the preset voltage value, it means that the voltage between the battery cells is close to being balanced. At this time, it is suitable to use the secondary balancing circuit for fine adjustment to ensure that the balancing process is more stable and efficient. Through this decision-making process, the type of the target balancing circuit is determined. The target duty cycle value sequence is generated according to the balancing mode. Since the operating modes of the DC-DC converter and the Buck-Boost converter both rely on the PWM (pulse width modulation) signal, the appropriate duty cycle is calculated to ensure that the converter operates in the required balancing mode. For the constant current mode of the DC-DC converter, the duty cycle calculation is adjusted according to the input voltage and the target balancing current to maintain a constant output current. For the constant voltage mode of the Buck-Boost converter, the duty cycle is calculated and dynamically adjusted based on the input voltage and the target output voltage to ensure the stability of the output voltage. After the target duty cycle value sequence is calculated, the PWM signal is modulated to generate the drive signal of the switch tube. When the target balancing loop is determined to be the main balancing loop, the generated switch tube drive signal is input into the main switch tube of the DC-DC converter, and the constant current output is achieved by accurately controlling the on-off timing of the main switch tube. The constant current control of the DC-DC converter adopts a current closed-loop feedback mechanism to ensure that the output current can be stably maintained at the set value. By real-time monitoring of the output current and dynamically adjusting the PWM duty cycle, the impact of load changes on the balancing current is effectively compensated to ensure the stability and reliability of the balancing process. When the target balancing loop is determined to be the secondary balancing loop, the generated switch tube drive signal is input into the four switch tubes of the Buck-Boost converter, and the on-off timing of the four switch tubes is accurately controlled to achieve constant voltage output. The constant voltage mode of the Buck-Boost converter adopts a voltage feedback control method, that is, by detecting the output voltage in real time and comparing it with the target voltage, the PWM duty cycle is adjusted so that the output voltage can be stably maintained at the set value. Since the Buck-Boost converter has the ability to step up and down, it can achieve a stable output voltage under different input voltage conditions, so that it can more flexibly adapt to the state of different battery cells during the balanced charging process. During the balancing process, the junction temperature of the switch tube in the target balancing loop is detected in real time to prevent the switch tube from being damaged due to overheating. The junction temperature of the switch tube is monitored by a thermistor or infrared temperature sensor, and the temperature data is input into the control system for real-time analysis.When it is detected that the junction temperature of the switch tube exceeds the target temperature, the target duty cycle value sequence is derated according to the temperature over-limit value to reduce the workload of the switch tube and reduce heat loss to obtain the first duty cycle value sequence. The derated calculation method adopts a linear attenuation or exponential attenuation method, that is, as the junction temperature rises, the PWM duty cycle is gradually reduced to reduce power loss and reduce the temperature of the switch tube, thereby ensuring the safe operation of the balancing loop. The first duty cycle value sequence is input into the steady-state duty cycle compensator to feedback compensate the derated duty cycle value. The steady-state duty cycle compensator fine-tunes the duty cycle according to historical data and real-time working status to avoid the degradation of balancing performance caused by the derating calculation. The compensator adopts a PID (proportional-integral-differential) control algorithm or an adaptive control algorithm based on a neural network to ensure the accuracy and stability of the duty cycle adjustment. After the compensation is completed, the second duty cycle value sequence is finally obtained, and the switch tube drive signal is updated to optimize the balancing charging process, so that it can maximize the balancing efficiency and battery life while ensuring safety.

[0023] The standard deviation of the terminal voltage value is calculated to calculate the voltage standard deviation, which is used to measure the voltage consistency of each single cell in the power battery pack. A larger standard deviation indicates that there is a larger voltage deviation in the battery pack, so a more active balancing strategy is needed to reduce this unbalanced state. At the same time, the maximum and minimum values ​​of the temperature value are calculated to calculate the temperature distribution range, which is used to measure the uniformity of the temperature distribution inside the power battery pack. If the temperature distribution range is large, there will be a local overheating problem, which will affect the battery performance and even bring safety hazards. During the balanced charging process, the charging strategy needs to be adjusted appropriately to reduce the temperature difference inside the battery pack. The root mean square deviation of the state of charge value is calculated to obtain the consistency index, which is used to evaluate the overall uniformity of the battery pack SOC (state of charge). The imbalance of SOC will cause some batteries to be overcharged or over-discharged, thereby affecting the service life of the battery. After completing the calculation of the above indicators, the voltage standard deviation, temperature distribution range and consistency index are weighted and combined to generate a reward function value, which is used to guide the Actor-Critic network to optimize the balanced charging strategy. In the reinforcement learning framework, the reward function is the core of the optimization goal. Its design needs to ensure that the charging strategy develops in the direction of reducing battery pack imbalance, improving energy utilization efficiency and extending battery life. After calculating the reward function value, it is subtracted from the target reward value to obtain the advantage function value, which is used to measure the improvement space of the current charging strategy compared to the optimal strategy. The larger the advantage function value, the closer the current strategy is to the optimal solution. Conversely, the balanced charging strategy needs to be further optimized. The policy gradient is calculated based on the advantage function value, and the policy gradient is truncated to prevent gradient explosion or gradient disappearance problems. The purpose of policy gradient truncation is to limit the amplitude of gradient changes, make the training process more stable, and avoid excessive updates of the Actor network due to extreme conditions, which affects convergence. After obtaining the truncated policy gradient, the policy entropy is calculated according to the logarithmic probability density of the output action of the Actor network in the action space. The policy entropy is used to measure the exploratory nature of the policy. A larger policy entropy means that the balanced charging strategy still maintains a certain degree of randomness under different conditions, thereby exploring a better balancing solution, while a smaller policy entropy indicates that the balanced charging strategy tends to converge and is more inclined to use the learned balancing mode. In the optimization process, the policy gradient and policy entropy are comprehensively considered to achieve a balance between exploration and utilization. The truncated policy gradient and policy entropy are weighted and superimposed, and the weight parameters of the Actor network are updated according to the set learning rate to obtain the updated Actor network weight parameters. The update of the Actor network can adjust the output of the balancing current, balancing mode, and balancing time, so that the balancing strategy can more accurately adapt to different battery states, thereby improving the effect of balanced charging.While optimizing the Actor network, the time difference error is calculated according to the advantage function value, and the mean square error of the time difference error is calculated to obtain the value loss function. The calculation of the value loss function is the key to updating the Critic network. Its goal is to minimize the error between the predicted value and the actual reward, thereby improving the value evaluation ability of the Critic network for the balanced charging strategy. Based on the value loss function, the weight parameters of the Critic network are updated according to the learning rate to obtain the updated Critic network weight parameters. According to the updated Actor network weight parameters and the updated Critic network weight parameters, the balanced current value, balanced mode and balanced time are corrected to obtain the optimized balanced charging strategy. Through continuous iterative optimization, the balanced charging strategy is dynamically adjusted under different working conditions to minimize the imbalance state inside the power battery pack, improve energy utilization efficiency, and extend the service life of the battery pack.

[0024] Data collection is performed on the cell voltage, temperature and balancing current in the main balancing loop and the secondary balancing loop, and a cell monitoring data matrix is ​​constructed to reflect the operating status of the power battery pack during the balancing charging process. During the data collection process, the terminal voltage of each cell is measured in real time by a high-precision voltage sensor, and its change over time is recorded. At the same time, the temperature sensor is used to obtain the temperature data of each cell to monitor the temperature anomaly during the charging process. The balancing current in the balancing loop is measured, and high-precision current data is obtained through a Hall sensor or a shunt resistor to ensure the stability and safety of the balancing process. After the construction of the cell monitoring data matrix is ​​completed, the cell voltage is compared with the preset voltage range to determine whether there is a voltage anomaly in the battery cell. When the voltage of a battery cell exceeds the preset range, a voltage over-limit mark is generated so that appropriate protection measures can be taken for the battery cell in the subsequent control process. The temperature difference between adjacent cells is calculated and compared with the preset temperature difference threshold to identify local overheating. If the temperature difference between a battery cell and the temperature between adjacent cells exceeds the set threshold, a temperature difference over-limit mark is generated to ensure that there is no safety hazard caused by overheating during the charging process. At the same time, the current threshold of the balancing current is compared with the rated current to determine whether the balancing current exceeds the safety range, and a current over-limit mark is generated accordingly. An over-limit level matrix is ​​constructed according to the voltage over-limit mark, the temperature difference over-limit mark and the current over-limit mark, and the matrix is ​​prioritized to generate the corresponding protection action instruction. In the process of generating the protection action instruction, different types of over-limit situations are weighted to determine the severity of various abnormalities, and the corresponding processing strategy is set accordingly. For example, if the voltage of a battery cell deviates from the normal range but does not exceed the safety limit, priority is given to reducing the balancing current to alleviate the voltage difference, and if the battery temperature exceeds the set safety limit, the balancing charge is immediately interrupted to prevent overheating and damage to the battery. When the protection action instruction is to reduce the balancing current, the current balancing current value is calculated in a decreasing manner to obtain the balancing current value after de-rating, and it is updated to the main balancing circuit or the secondary balancing circuit to ensure that the balancing charging process can be carried out within the safety range. The derating calculation of the balancing current adopts a linear decreasing or exponential decreasing strategy, that is, the current is gradually reduced while maintaining the balancing effect to reduce the excessive voltage difference between battery cells and prevent overcurrent from damaging the battery. During the equalization current adjustment process, dynamic adjustment is performed in combination with the overall state of the battery pack to ensure that the equalization process can maintain the equalization effect without causing additional negative impacts on the battery pack. When the protection action instruction is to interrupt the equalization charge, an equalization interruption signal sequence is immediately generated and the signal sequence is input into the switch control terminals of the main equalization loop and the secondary equalization loop respectively, so that the corresponding switch tube enters the cut-off state, thereby stopping the output of the equalization current.The purpose of equalization interruption is to prevent the abnormal situation from deteriorating continuously. For example, when the battery temperature reaches the critical point or the voltage of a single battery greatly deviates from the normal value, the equalization charging is interrupted to effectively avoid further unbalanced state and provide time for subsequent fault recovery. During the equalization interruption process, ensure the rapid response of the control signal so as to cut off the equalization circuit in time and prevent the abnormal state from causing irreversible damage to the battery pack. Feedback detection is performed on the equalization current value after derating or the execution status of the equalization interruption signal sequence to ensure that the protection measures are correctly executed. The feedback detection content includes whether the equalization current is reduced as expected, whether the switch tube of the equalization circuit successfully enters the cut-off state, and whether the voltage and temperature of the battery cell return to the safe range. Through these feedback data, the equalization charging strategy is optimized, and the protection measures are adjusted when necessary to ensure the stability and safety of the charging process. After the protection action is completed, the protection execution results are recorded in the equalization charging log for subsequent analysis and maintenance. The records of the equalization charging log include the voltage and temperature change trends of the battery cells, the adjustment history of the equalization current, the occurrence time of the abnormal state and the treatment measures. Through the analysis of the equalization charging log, the equalization strategy is optimized, the service life of the power battery pack is improved, and the safety risks in the charging process are reduced.

[0025] In the embodiment of the present invention, by introducing a multi-head self-attention network for feature extraction, efficient processing of power battery pack state information is achieved, the correlation between different monomers is accurately captured, and the expression ability of state features is improved. The Actor-Critic deep reinforcement learning framework is adopted to dynamically generate a combination strategy of balancing current, balancing mode and balancing time, and adaptive adjustment of balancing control is achieved, so that the balancing strategy can be optimized according to the real-time state of the battery pack. A multi-level balancing circuit structure in which the main balancing loop and the secondary balancing loop work together is designed, and a balancing mode switching mechanism based on voltage difference is established, which reduces energy loss while ensuring balancing efficiency. A complete balancing charging protection system is constructed, and the safety and reliability of the balancing charging process are ensured by real-time monitoring of monomer voltage, temperature and balancing current, and setting multi-level protection thresholds. The proximal strategy optimization algorithm is adopted to update the network parameters online, and the exploration efficiency of reinforcement learning is improved by introducing advantage functions and strategy entropy, and the convergence performance of the algorithm is enhanced. The temperature adaptive control of the balancing circuit is realized, and the overheating of the device is effectively prevented through real-time monitoring of the junction temperature of the switch tube and dynamic adjustment of the duty cycle, which prolongs the service life of the system. Through the comprehensive evaluation of voltage standard deviation, temperature distribution extreme difference and charge state consistency indicators, quantitative evaluation and strategy optimization of balancing effect are achieved.

[0026] In a specific embodiment, the process of executing step S100 may specifically include the following steps: Setting sampling timing for multiple sampling points in the power battery pack to obtain a parameter acquisition timing matrix; According to the parameter acquisition timing matrix, voltage data is collected at the sampling points to obtain the terminal voltage value, temperature data is collected at the sampling points to obtain the temperature value, and current data is collected at the sampling points to obtain the charge and discharge current value according to the parameter acquisition timing matrix; Perform time integration operation on the charge and discharge current value and superimpose it with the initial state of charge value to obtain a target value, and standardize the target value with the nominal capacity of the battery to obtain a state of charge value; The terminal voltage value and the charge and discharge current value are differentially calculated at adjacent sampling points to obtain a differential result, and the differential result is subjected to Kalman filtering and ratio calculation to obtain the internal resistance change rate.

[0027] Specifically, a reasonable sampling sequence is defined to ensure that the voltage, temperature, and charge and discharge current of each single cell are measured under a unified time reference. The sampling period is set to , then in each sampling period, the voltage, temperature and current of all single cells are measured in turn, and the collected data are arranged in chronological order to form a parameter collection timing matrix. The battery in The voltage, temperature and current in a sampling period are , and , then the parameter acquisition timing matrix is ​​expressed as: in, Representative in the The measurement data matrix of all single cells in a sampling period. The charge and discharge current values ​​are time-integrated and superimposed with the initial state of charge value to calculate the state of charge of the power battery. The calculation of SOC is based on the Coulomb integration method, and the mathematical expression is: in, Representative The state of charge within a sampling period; is the initial state of charge, provided by the manufacturer or estimated based on the open circuit voltage; Represents the nominal capacity of the battery; For the The battery current within a sampling period; Represents the sampling period. After calculating the state of charge, the terminal voltage value and the charge and discharge current value are differentially calculated at adjacent sampling points to calculate the internal resistance change rate. Assume that In adjacent sampling cycles, and The voltage and current are , , and , then the difference is calculated as follows: After completing the differential calculation, the internal resistance is calculated using Ohm's law: Kalman filtering is performed on the voltage differential and current differential to improve the measurement accuracy. The update equation of Kalman filtering is as follows: in, Represents the estimated value of internal resistance after filtering; is the Kalman gain, which is used to adjust the sensitivity of the estimated value to the new measurement value; is the internal resistance value calculated currently. After filtering, a more stable internal resistance change rate is obtained, which can be used to evaluate the battery health status and optimize the balancing strategy.

[0028] In a specific embodiment, the process of executing step S200 may specifically include the following steps: Perform feature splicing and dimension transformation on the terminal voltage value, temperature value, charge and discharge current value, state of charge value and internal resistance change rate to obtain a state feature sequence; Perform linear transformation on the state feature sequence to generate query matrix, key matrix and value matrix, and perform matrix multiplication operation on the query matrix and the key matrix to obtain the attention score matrix; Perform Softmax normalization on the attention score matrix to obtain the attention weight matrix, and perform matrix multiplication on the attention weight matrix and the value matrix to obtain the single-head attention output feature; The state feature sequences are input into N parallel attention calculation units in the multi-head self-attention network. Each attention calculation unit independently performs attention feature extraction to obtain N attention output features. The N attention output features are concatenated to obtain the target feature vector, and the target feature vector is normalized by the LayerNorm layer to obtain the state-related feature matrix.

[0029] Specifically, the terminal voltage value, temperature value, charge and discharge current value, state of charge value and internal resistance change rate are subjected to feature splicing and dimension transformation. Assume that the power battery pack contains single cell, and in the At each sampling moment, the terminal voltage, temperature, charge and discharge current, state of charge and internal resistance change rate of each battery are expressed as , , , and , the state characteristic matrix of the entire battery pack is expressed as: in, The rows represent different single cells, and the columns represent different features. Perform feature concatenation and dimension transformation. Assume that the feature vector Indicates Battery status: The state characteristic sequence of the entire battery pack is expressed as: in The dimension is ,in Represents the number of features. Perform linear transformation on the state feature sequence to generate the query matrix (Query), key matrix (Key) and value matrix (Value). , and are trainable weight matrices used to project feature sequences to compute queries, keys, and values, respectively: in, is the query matrix, which represents the importance evaluation of the input features; is the key matrix, which represents the similarity between input features; is a value matrix, representing the information of the input feature itself. Calculate the attention score matrix , which is calculated by taking the query matrix and key matrix Perform a dot product operation and scale it: in, is the feature dimension, the scaling factor Used to stabilize the gradient and improve the stability of numerical calculations. Perform Softmax normalization operation to generate the attention weight matrix : Among them, the role of Softmax normalization is to convert the attention score into a probability distribution so that the sum of the weights of each input feature is 1, thereby ensuring that the attention mechanism effectively captures the importance of the input features. With value matrix Perform matrix multiplication to calculate the output features of a single-head attention: in, Represents the state feature output after attention weighting. The state feature sequence is input into the multi-head self-attention network parallel attention computing units, each unit performs attention feature extraction independently. The query, key, and value projection matrices corresponding to the attention heads are , and , which is calculated as: Calculate the weight of each attention head: And calculate the attention output of this head: all After the attention heads are calculated, their output features are concatenated: in is the output of the multi-head attention mechanism, and its dimension is . For the concatenated target feature vector Perform LayerNorm normalization to improve training stability: Among them, the role of LayerNorm normalization is to normalize the features within each sample, so that the model can maintain a good learning effect under different batches of data input. Get the state correlation feature matrix , contains status information such as voltage, temperature, current, state of charge and internal resistance change rate of each single cell in the power battery pack, and extracts the correlation between these features through the multi-head self-attention mechanism.

[0030] In a specific embodiment, the process of executing step S300 may specifically include the following steps: The state-related feature matrix is ​​input into the first fully connected layer of the Actor network. The first fully connected layer contains 512 neurons and is nonlinearly transformed through the ReLU activation function to obtain the first layer of action features. The first-layer action features are input into the second fully connected layer of the Actor network. The second fully connected layer contains 256 neurons and is nonlinearly transformed through the ReLU activation function to obtain the second-layer action features. The second-layer action features are input into the third fully connected layer of the Actor network. The third fully connected layer contains 32 neurons and is nonlinearly transformed through the Tanh activation function to obtain the action output features. The action output features are input into three parallel action mapping layers respectively, the first action mapping layer outputs the balanced current value, the second action mapping layer outputs the balanced mode of the constant current mode or the constant voltage mode, and the third action mapping layer outputs the balanced time; The state association feature matrix and the action output feature combination are input into two Q networks with the same structure. Each Q network contains three fully connected layers. The number of neurons in the three fully connected layers is 512, 256, and 1 respectively, to obtain the first Q value and the second Q value. The first Q value and the second Q value are minimum selected to obtain the target Q value, and the target Q value is used as the value evaluation indicator of the Actor network output action.

[0031] Specifically, a state-related characteristic matrix is ​​defined, which contains the key state parameters of the power battery pack, such as terminal voltage, temperature, charge and discharge current, state of charge, and internal resistance change rate. Assume that the power battery pack contains single cell, and in the At sampling moments, the state characteristics of each battery are expressed as ,in: in, For the The terminal voltage of a single cell, is the temperature, is the charge and discharge current, is the state of charge, is the internal resistance change rate. The state characteristic matrix of the entire power battery pack is expressed as: This matrix is ​​used as the input of the Actor network and enters the first fully connected layer, which contains 512 neurons and is nonlinearly transformed by the ReLU activation function. Suppose the input matrix After the first layer weight matrix and bias After the transformation, the first layer of action features is calculated: in, is the output feature matrix of the first layer, is the first layer weight matrix, The ReLU activation function is used as a bias term to introduce nonlinearity, which enables the model to learn complex state-action mapping relationships more effectively. Enter the second fully connected layer, which contains 256 neurons and also uses the ReLU activation function for nonlinear transformation. The calculation method is as follows: in, and are the weight matrix and bias term of the second layer respectively. The main function of this layer is to further extract more abstract action features so that the Actor network can more accurately fit the complex mapping relationship of the equilibrium strategy. Enter the third fully connected layer, which contains 32 neurons and uses the Tanh activation function for nonlinear transformation to generate the final action output features: Among them, the Tanh activation function is chosen because its output range is between [-1,1], which makes the output of the Actor network have good numerical stability, and Tanh has a large gradient near the zero point, which helps to improve learning efficiency. After generating the action output features, they are input into three parallel action mapping layers to calculate the balanced current value, balanced mode, and balanced time respectively. Among them, the output of the first action mapping layer corresponds to the balanced current value: in, Indicates the balancing current value, and are the weights and biases of the mapping layer respectively. The output of the second action mapping layer is used to determine the balancing mode, that is, to select between constant current mode and constant voltage mode: in, represents the probability value of the equilibrium mode, Represents the Sigmoid function, which is used to limit the output value between (0,1) so as to determine whether it is constant current mode or constant voltage mode according to the threshold. The output of the third action mapping layer is used to determine the balancing time: in, Represents the equilibrium time. Evaluate the actions generated by the Actor network and associate the state feature matrix and action output features After the combination, the two Q networks are input into the Critic network. Let the input matrix after the combination be: The matrix is ​​input into two Q networks with the same structure. Each Q network contains three fully connected layers with 512, 256 and 1 neurons respectively. For the first Q network, the calculation process is as follows: in, is the output value of the first Q network. Similarly, the calculation process of the second Q network is as follows: in, is the output value of the second Q network. After obtaining the first Q value and the second Q value, a minimum value selection operation is performed to reduce the problem of over-optimism of the strategy: in, As the final target Q value, it is used to evaluate the value of the Actor network output action. The Actor network continuously optimizes the balancing strategy based on the value evaluation of the Critic network, so that the balanced charging process dynamically adapts to different battery states and improves the efficiency and reliability of balanced charging.

[0032] In a specific embodiment, the process of executing step S400 may specifically include the following steps: Compare and analyze the balancing current value and the voltage difference of adjacent single cells. When the voltage difference is greater than the preset voltage value, select the main balancing circuit for balancing charging. When the voltage difference is less than or equal to the preset voltage value, select the secondary balancing circuit for balancing charging to obtain the target balancing circuit. According to the type and the balancing mode of the target balancing loop, a target duty cycle value sequence is generated, and pulse width modulation is performed on the target duty cycle value sequence to obtain a switch tube driving signal; When the target balancing loop is the main balancing loop, the switch tube driving signal is input into the main switch tube of the DC-DC converter to control the on-off timing of the main switch tube to obtain a constant current output; When the target balancing loop is a sub-balancing loop, the switch tube driving signal is input into the four switch tubes of the Buck-Boost converter to control the on-off timing of the four switch tubes to obtain a constant voltage output; The junction temperature of the switch tube in the target balancing loop is detected in real time. When the junction temperature of the switch tube exceeds the target temperature, the target duty cycle value sequence is derated according to the temperature over-limit value to obtain a first duty cycle value sequence; The first duty cycle value sequence is input into the steady-state duty cycle compensator, feedback compensation is performed on the first duty cycle value sequence to obtain a second duty cycle value sequence, and the switch tube drive signal is updated.

[0033] Specifically, the voltage difference between adjacent single cells is calculated and compared with a preset voltage threshold. A single cell, in the At the sampling moment, the terminal voltage of each single cell is , the voltage difference between adjacent single cells is calculated as follows: in, Indicates and The voltage difference of the batteries. Assume that the preset voltage threshold is ,like , it indicates that the voltage imbalance between batteries is large, and the main balancing circuit is selected for balanced charging; if , it indicates that the voltage deviation is small, and the secondary balancing circuit is selected for fine-tuning balancing. The selection rule of the target balancing circuit is expressed as: in, Representative in the The type of balancing loop selected at the sampling moment. When the balancing loop type is determined, the target duty cycle value sequence is generated according to the balancing mode (constant current mode or constant voltage mode), and a pulse width modulation (PWM) signal is generated based on this to control the output of the balancing current or balancing voltage. For the main balancing loop, the DC-DC converter output constant current mode is adopted, and the duty cycle The calculation method is: in, is the converter input voltage, is the target balancing voltage. For the secondary balancing loop, the Buck-Boost converter output constant voltage mode is adopted, and the duty cycle The calculation is as follows: The calculated target duty cycle is converted into a switch tube drive signal using PWM modulation. : in, is the PWM period, is the target duty cycle in the current sampling period. When the target balancing loop is the main balancing loop, the switch tube drive signal The main switch of the input DC-DC converter is used to control its on-off timing to achieve constant current output. The on-off behavior of the main switch is determined by the PWM signal, so that the current is balanced. Keep at set value: in, is the equivalent resistance of the balancing loop. When the target balancing loop is a secondary balancing loop, the four switches of the Buck-Boost converter are controlled simultaneously to switch synchronously according to the PWM signal to keep the balanced output voltage stable. The switching logic of the four switches is based on the duty cycle sequence , so that the balanced voltage output by the Buck-Boost converter meets the target requirements: During the balancing process, the junction temperature of the switch tube in the target balancing loop is detected in real time to prevent the temperature from exceeding the limit and causing damage to the device. Assume that the current junction temperature of a switch tube is , the temperature threshold is set to , then when When the target duty cycle is derated, the target duty cycle is derated to reduce power loss. The derated calculation method uses linear attenuation: in, is the maximum allowable temperature, is the duty cycle value after derating. Enter the steady-state duty cycle compensator to dynamically adjust it to ensure that the system can still operate stably under different temperature conditions. The compensator is based on the PID control algorithm for feedback regulation: ; in, are the proportional, integral and differential coefficients of PID control respectively, is the duty cycle of the previous cycle. The duty cycle after compensation Used to update the switch tube drive signal To ensure that the balancing circuit works normally within the temperature controlled range and maintain the best balancing charging effect.

[0034] In this embodiment, controlling the DC-DC converter in the main balancing loop to generate a constant current output or controlling the Buck-Boost converter in the secondary balancing loop to generate a constant voltage output also includes: performing difference calculation on the actual balancing current of the main balancing loop and the secondary balancing loop and the target balancing current to obtain a balancing current deviation sequence, constructing a Markov random jump model based on the balancing current deviation sequence, and obtaining a fault feature matrix; constructing a generalized Martingale measure based on the fault feature matrix, performing a random walk analysis on the balancing current deviation sequence to obtain a fault probability, and grading the balancing loop according to the fault probability to obtain a fault level index; inputting the fault level index into a fault-tolerant equivalent controller to generate a fault compensation control quantity, the fault-tolerant equivalent controller adopts a three-layer neural network structure, including an input layer, a hidden layer and an output layer, and the hidden layer adopts a radial basis function as an activation function; combining the fault compensation control quantity with the balancing current value to obtain a corrected The balanced current value is obtained, and the PWM control signal is regenerated according to the corrected balanced current value to obtain the compensated switch tube control signal; the control effect of the compensated switch tube control signal is evaluated online, the probability distribution difference between the actual balanced current and the target balanced current is calculated, the control performance index is obtained, and the network parameters of the fault-tolerant equivalent controller are adaptively updated according to the control performance index; the probability limit constraint condition is constructed, the fault influence coefficient is calculated according to the balanced current deviation sequence, the fault influence coefficient is compared with the preset threshold, and the backup balancing loop is triggered when the preset threshold is exceeded; based on the working status of the backup balancing loop, the compensation strategy of the fault-tolerant equivalent controller is dynamically adjusted to generate a new fault compensation control quantity, and the new fault compensation control quantity is fed back to the balanced control system; the compensated switch tube control signal, fault feature matrix and control performance index are recorded in the fault diagnosis database for subsequent fault mode analysis and control strategy optimization.

[0035] In a specific embodiment, the process of executing step S500 may specifically include the following steps: Perform standard deviation calculation on the terminal voltage value to obtain the voltage standard deviation, perform maximum and minimum calculation on the temperature value to obtain the temperature distribution range, and perform root mean square deviation calculation on the state of charge value to obtain the consistency index; The voltage standard deviation, temperature distribution range and consistency index are weighted and combined to generate the reward function value, and the difference between the reward function value in the current state and the target reward value is calculated to obtain the advantage function value; The policy gradient is calculated based on the advantage function value, and the policy gradient is truncated to obtain the truncated policy gradient. The policy entropy value is calculated based on the logarithmic probability density of the output action of the Actor network in the action space. Perform weighted superposition on the truncated policy gradient and policy entropy value, update the weight parameters of the Actor network according to the learning rate, and obtain the updated weight parameters of the Actor network; The time series difference error is calculated according to the advantage function value, and the mean square error of the time series difference error is calculated to obtain the value loss function. The weight parameters of the Critic network are updated according to the learning rate to obtain the updated weight parameters of the Critic network. According to the updated Actor network weight parameters and the updated Critic network weight parameters, the balancing current value, balancing mode and balancing time are corrected to obtain the optimized balancing charging strategy.

[0036] Specifically, the standard deviation of the terminal voltage is calculated to measure the consistency of the voltage in the battery pack. A single cell, in the During the sampling period, the terminal voltage of each single cell is , then the voltage standard deviation The calculation is as follows: in, is the average voltage of the battery pack: The larger the voltage standard deviation, the more serious the imbalance of the battery pack, and a stronger balancing strategy is needed to adjust the voltage difference of the single battery. Calculate the temperature distribution range to measure the temperature non-uniformity inside the battery pack. A single cell in The temperature at the moment is , the temperature distribution is extremely poor The calculation is as follows: The greater the temperature distribution range, the more serious the local overheating problem is inside the battery pack. At this time, the balanced charging strategy needs to be adjusted to reduce the heat accumulation in the high-temperature area. In order to evaluate the balance of the state of charge (SOC), the consistency index of SOC is calculated, and the root mean square deviation is used to measure the consistency of SOC. Assume that the SOC of each single cell is , then the SOC consistency index The calculation is as follows: in, For the average SOC: The smaller the SOC consistency index is, the more balanced the SOC state in the battery pack is, and the more reasonable the energy distribution during charging and discharging is. The voltage standard deviation, temperature distribution extremes, and SOC consistency index are weighted together to construct a reward function for balanced charging. Let the reward function be The weighted sum of these three indicators is: in, is the weight coefficient, which measures the relative importance of voltage balance, temperature balance and SOC balance respectively. The goal of the reward function is to minimize the degree of imbalance, so it takes a negative sign, so that the better the balance state, the higher the reward value. The larger the value, the better. Calculate the advantage function value , which is used to measure the improvement of the current strategy relative to the baseline strategy. Let the target reward value be , then the advantage function is calculated as follows: if If it is positive, it means that the current strategy is better than the benchmark strategy and needs to be strengthened; if If it is negative, it means that the current strategy is not good and needs to be adjusted. Based on the advantage function, the policy gradient is calculated and truncated to prevent the gradient explosion problem. Suppose the policy gradient of the Actor network is : in, is the action probability density output by the policy network, are the parameters of the Actor network. In order to prevent the gradient from being too large and causing unstable training, the policy gradient is truncated: in, is the gradient cutoff threshold. Calculate the policy entropy of the Actor network to encourage the exploration of the policy. The policy entropy is calculated as follows: The larger the policy entropy, the more exploratory the Actor network is in the balanced charging strategy, which helps to improve the robustness of the charging strategy. and policy entropy Perform weighted superposition and follow the learning rate Update the parameters of the Actor network: in, is the policy entropy weight, which is used to control the impact of exploration on policy optimization. At the same time, the Critic network needs to be updated to optimize the state value estimation. The goal of the Critic network is to minimize the temporal difference error (TD error), which is calculated as follows: in, is the state value estimated by the Critic network, is the discount factor. The loss function of the Critic network is calculated using the mean square error: According to the learning rate Update the parameters of the Critic network : After updating the parameters of the Actor and Critic networks, the balancing current, balancing mode and balancing time are corrected according to the optimized Actor network output to optimize the balancing charging strategy. The corrected balancing charging strategy dynamically adjusts the balancing charging parameters according to the current battery status, thereby improving the balance of the charging process, reducing the imbalance within the battery pack, increasing battery life, and optimizing energy utilization efficiency.

[0037] In a specific embodiment, the process of executing step S600 may specifically include the following steps: Collect data on the cell voltage, temperature and balancing current in the main balancing loop and the secondary balancing loop to obtain a cell monitoring data matrix; Compare the cell voltage in the cell monitoring data matrix with the preset voltage range to obtain a voltage over-limit mark, compare the temperature difference between adjacent cells with the preset temperature difference threshold to obtain a temperature difference over-limit mark, and compare the balanced current with the current threshold of the rated current to obtain a current over-limit mark; Generate an over-limit level matrix according to the voltage over-limit mark, the temperature difference over-limit mark and the current over-limit mark, and prioritize the over-limit level matrix to obtain a protection action instruction; When the protection action instruction is to reduce the balancing current, the balancing current value is calculated downward to obtain the balancing current value after de-rating, and the balancing current value after de-rating is updated to the main balancing circuit or the secondary balancing circuit; When the protection action instruction is to interrupt the balanced charging, a balanced interrupt signal sequence is generated, and the balanced interrupt signal sequence is input into the switch tube control terminals of the main balanced circuit and the secondary balanced circuit respectively, so that the switch tube enters the cut-off state; Feedback detection is performed on the execution status of the balancing current value or the balancing interrupt signal sequence after derating to obtain the protection execution result, and the protection execution result is recorded in the balancing charging log.

[0038] Specifically, a high-precision sensor is arranged inside the power battery pack to monitor the working status of each single battery in real time. Single cell, at the sampling time , the terminal voltage, temperature and balancing current of each single cell are expressed as , and ,in . Through the data acquisition system, construct a single monitoring data matrix: Each row corresponds to the monitoring data of a single cell, and each column represents a different measurement parameter. Comparison operations are performed to determine whether the voltage exceeds the limit. For each single battery, if Out of range: Then the battery voltage is marked as out of limit: otherwise: in, Representative The voltage of each battery is over the limit. The temperature difference between adjacent single batteries Calculate and compare with the preset temperature difference threshold For comparison: if: Then the battery temperature difference is marked as out of limit: otherwise: in, Representative The temperature difference of each battery exceeds the limit mark. Similarly, the balancing current With rated current Compare to determine whether the balancing current exceeds the limit: If the balancing current exceeds the threshold, then: otherwise: After calculating the voltage over-limit mark, temperature difference over-limit mark and current over-limit mark, combine them into an over-limit grade matrix: Prioritize the over-limit level matrix to determine the triggering order of the protection action instructions. According to the sorting results, the corresponding protection action instructions are obtained. If the protection action instruction requires reducing the balancing current, the current balancing current is Perform a decrement calculation: in, is the derating ratio coefficient, which is taken as The derating balancing current is updated to the main balancing circuit or the secondary balancing circuit. If the protection action instruction requires interruption of balancing charging, a balancing interruption signal sequence is generated and input into the switch control terminal of the main balancing circuit and the secondary balancing circuit to make the switch enter the cut-off state: in, Representative The switch status of each battery, 0 means off, 1 means on. After the protection action is executed, feedback detection is performed on the derating balancing current value or the execution status of the balancing interrupt signal sequence to confirm whether the protection measures are effective. If the feedback detection shows that the balancing current has been reduced to the set range or the balancing circuit has been interrupted, the execution is recorded as successful: Otherwise, the execution fails: The protection execution results are recorded in the balanced charging log for subsequent analysis and optimization of the balanced strategy. The balanced charging log records are as follows: in, Record the current monitoring data, over-limit status, adjusted balancing current, switch status and execution results.

[0039] See also Figure 2 , Figure 2 A schematic block diagram of the structure of a balanced charging device 200 for a power battery pack provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the balanced charging device 200 of the power battery pack includes: The parameter acquisition module 210 is used to acquire parameters of the power battery pack, measure the terminal voltage value, temperature value and charge and discharge current value of each single battery, and calculate the state of charge value and internal resistance change rate of each single battery; The feature extraction module 220 is used to construct the terminal voltage value, temperature value, charge and discharge current value, charge state value and internal resistance change rate into a state feature sequence, and input it into the multi-head self-attention network for feature extraction to generate a state correlation feature matrix; A construction module 230 is used to construct a three-layer fully connected Actor network and a Critic network with a dual-Q structure based on the state association feature matrix, and output a balanced current value, a constant current or constant voltage balanced mode, and a balanced time through the Actor network; The control module 240 is used to control the DC-DC converter in the main balancing loop to generate a constant current output, or control the Buck-Boost converter in the secondary balancing loop to generate a constant voltage output according to the balancing current value, balancing mode and balancing time.

[0040] Through the collaboration of the above components and the introduction of multi-head self-attention network for feature extraction, the efficient processing of power battery pack state information is achieved, the correlation between different monomers is accurately captured, and the expression ability of state features is improved. The Actor-Critic deep reinforcement learning framework is adopted to dynamically generate the combination strategy of balancing current, balancing mode and balancing time, realize the adaptive adjustment of balancing control, and enable the balancing strategy to be optimized according to the real-time state of the battery pack. A multi-level balancing circuit structure with the main balancing loop and the secondary balancing loop working together is designed, and a balancing mode switching mechanism based on voltage difference is established, which reduces energy loss while ensuring balancing efficiency. A complete balancing charging protection system is constructed, which ensures the safety and reliability of the balancing charging process by real-time monitoring of monomer voltage, temperature and balancing current, and setting multi-level protection thresholds. The proximal strategy optimization algorithm is used to update the network parameters online. By introducing the advantage function and strategy entropy, the exploration efficiency of reinforcement learning is improved and the convergence performance of the algorithm is enhanced. The temperature adaptive control of the balancing circuit is realized. Through real-time monitoring of the junction temperature of the switch tube and dynamic adjustment of the duty cycle, the device is effectively prevented from overheating and the service life of the system is extended. Through the comprehensive evaluation of voltage standard deviation, temperature distribution extreme difference and charge state consistency indicators, quantitative evaluation and strategy optimization of balancing effect are achieved.

[0041] See also Figure 3 , Figure 3 The present invention is a schematic block diagram of the structure of a balanced charging device 300 for a power battery pack provided in an embodiment of the present application. The balanced charging device 300 for a power battery pack includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected via a device bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0042] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned balanced charging methods for the power battery pack.

[0043] The processor 301 is used to provide computing and control capabilities to support the operation of the equalizing charging device 300 of the entire power battery pack.

[0044] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned balanced charging methods for the power battery pack.

[0045] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the equalizing charging device 300 of the power battery pack involved in the solution of the present application. The specific equalizing charging device 300 of the power battery pack may include more or less components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0046] It should be understood that the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0047] It should be noted that those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the equalizing charging device 300 of the power battery pack described above can refer to the corresponding process of the equalizing charging method of the power battery pack described above, and will not be repeated here.

[0048] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0049] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0050] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A balanced charging method for a power battery pack, characterized in that: include: Collect parameters of the power battery pack, measure the terminal voltage, temperature, and charge and discharge current of each single battery, and calculate the state of charge and internal resistance change rate of each single battery; The terminal voltage value, the temperature value, the charge and discharge current value, the state of charge value and the internal resistance change rate are constructed into a state feature sequence, and input into a multi-head self-attention network for feature extraction to generate a state correlation feature matrix; Based on the state association feature matrix, a three-layer fully connected Actor network and a dual-Q structure Critic network are constructed, and a balanced current value, a constant current or constant voltage balanced mode and a balanced time are output through the Actor network; According to the balancing current value, the balancing mode and the balancing time, the DC-DC converter in the main balancing loop is controlled to generate a constant current output, or the Buck-Boost converter in the secondary balancing loop is controlled to generate a constant voltage output.

2. The balanced charging method for a power battery pack according to claim 1, characterized in that: The parameter collection of the power battery pack, measuring the terminal voltage value, temperature value and charge and discharge current value of each single battery, and calculating the state of charge value and internal resistance change rate of each single battery, includes: Setting a sampling sequence for a plurality of sampling points in the power battery pack to obtain a parameter acquisition sequence matrix; According to the parameter acquisition timing matrix, voltage data is collected at the sampling points to obtain terminal voltage values, according to the parameter acquisition timing matrix, temperature data is collected at the sampling points to obtain temperature values, and according to the parameter acquisition timing matrix, current data is collected at the sampling points to obtain charge and discharge current values; Performing a time integration operation on the charge and discharge current value and superimposing it with the initial state of charge value to obtain a target value, and normalizing the target value with the nominal capacity of the battery to obtain a state of charge value; A differential operation is performed on the terminal voltage value and the charge and discharge current value at adjacent sampling points to obtain a differential result, and the differential result is subjected to Kalman filtering and ratio calculation to obtain an internal resistance change rate.

3. The balanced charging method for a power battery pack according to claim 1, characterized in that: The terminal voltage value, the temperature value, the charge and discharge current value, the state of charge value and the internal resistance change rate are constructed into a state feature sequence, and input into a multi-head self-attention network for feature extraction to generate a state-related feature matrix, including: Performing feature splicing and dimension transformation on the terminal voltage value, the temperature value, the charge and discharge current value, the state of charge value, and the internal resistance change rate to obtain a state feature sequence; Performing a linear transformation on the state feature sequence to generate a query matrix, a key matrix and a value matrix, and performing a matrix multiplication operation on the query matrix and the key matrix to obtain an attention score matrix; Performing a Softmax normalization operation on the attention score matrix to obtain an attention weight matrix, and performing a matrix multiplication operation on the attention weight matrix and the value matrix to obtain a single-head attention output feature; Inputting the state feature sequences into N parallel attention calculation units in the multi-head self-attention network respectively, each attention calculation unit independently performs attention feature extraction to obtain N attention output features; The N attention output features are concatenated to obtain a target feature vector, and the target feature vector is normalized by the LayerNorm layer to obtain a state-related feature matrix.

4. The balanced charging method for a power battery pack according to claim 1, characterized in that: The method of constructing a three-layer fully connected Actor network and a dual-Q structured Critic network based on the state association feature matrix, and outputting a balanced current value, a constant current or constant voltage balanced mode, and a balanced time through the Actor network, includes: The state association feature matrix is ​​input into the first fully connected layer of the Actor network, where the first fully connected layer includes 512 neurons, and a nonlinear transformation is performed through a ReLU activation function to obtain the first layer of action features; The first layer of action features is input into the second fully connected layer of the Actor network. The second fully connected layer contains 256 neurons and is nonlinearly transformed through the ReLU activation function to obtain the second layer of action features. The second layer of action features is input into the third fully connected layer of the Actor network, where the third fully connected layer includes 32 neurons, and a nonlinear transformation is performed through a Tanh activation function to obtain action output features; The action output characteristics are input into three parallel action mapping layers respectively, the first action mapping layer outputs a balanced current value, the second action mapping layer outputs a balanced mode of a constant current mode or a constant voltage mode, and the third action mapping layer outputs a balanced time; Inputting the state association feature matrix and the action output feature combination into two Q networks with the same structure, each Q network comprises three fully connected layers, and the number of neurons in the three fully connected layers is 512, 256, and 1 respectively, to obtain a first Q value and a second Q value; A minimum value selection operation is performed on the first Q value and the second Q value to obtain a target Q value, and the target Q value is used as a value evaluation indicator of the Actor network output action.

5. The balanced charging method for a power battery pack according to claim 1, characterized in that: The controlling of the DC-DC converter in the main balancing loop to generate a constant current output, or the controlling of the Buck-Boost converter in the secondary balancing loop to generate a constant voltage output according to the balancing current value, the balancing mode and the balancing time, comprises: Compare and analyze the balancing current value and the voltage difference of adjacent single cells, select the main balancing circuit for balancing charging when the voltage difference is greater than a preset voltage value, and select the secondary balancing circuit for balancing charging when the voltage difference is less than or equal to the preset voltage value, to obtain a target balancing circuit; Generate a target duty cycle value sequence according to the type of the target balancing loop and the balancing mode, and perform pulse width modulation on the target duty cycle value sequence to obtain a switch tube drive signal; When the target balancing loop is the main balancing loop, the switch tube driving signal is input into the main switch tube of the DC-DC converter to control the on-off timing of the main switch tube to obtain a constant current output; When the target balancing loop is the secondary balancing loop, the switch tube driving signal is input into the four switch tubes of the Buck-Boost converter to control the on-off timing of the four switch tubes to obtain a constant voltage output; Performing real-time detection on the junction temperature of the switch tube in the target balancing loop, and when the junction temperature of the switch tube exceeds the target temperature, performing derating calculation on the target duty cycle value sequence according to the temperature over-limit value to obtain a first duty cycle value sequence; The first duty cycle value sequence is input into a steady-state duty cycle compensator, feedback compensation is performed on the first duty cycle value sequence to obtain a second duty cycle value sequence, and the switch tube drive signal is updated.

6. The balanced charging method for a power battery pack according to claim 1, characterized in that: The balanced charging method of the power battery pack also includes: Performing standard deviation calculation on the terminal voltage value to obtain the voltage standard deviation, performing maximum and minimum value calculation on the temperature value to obtain the temperature distribution range, and performing root mean square deviation calculation on the state of charge value to obtain the consistency index; The voltage standard deviation, the temperature distribution extreme difference and the consistency index are weightedly combined to generate a reward function value, and a difference operation is performed between the reward function value in the current state and the target reward value to obtain an advantage function value; Calculating a policy gradient based on the advantage function value, truncating the policy gradient to obtain a truncated policy gradient, and calculating a policy entropy value according to a logarithmic probability density of an output action of the Actor network in an action space; Performing weighted superposition on the truncated policy gradient and the policy entropy value, and updating the weight parameters of the Actor network according to the learning rate to obtain updated weight parameters of the Actor network; Calculating the time series difference error according to the advantage function value, and performing mean square error calculation on the time series difference error to obtain the value loss function, and updating the weight parameters of the Critic network according to the learning rate to obtain the updated weight parameters of the Critic network; According to the updated Actor network weight parameter and the updated Critic network weight parameter, the balancing current value, the balancing mode and the balancing time are corrected to obtain an optimized balancing charging strategy.

7. The balanced charging method for a power battery pack according to claim 1, characterized in that: The balanced charging method of the power battery pack also includes: Collecting data on the cell voltage, temperature and balancing current in the primary balancing loop and the secondary balancing loop to obtain a cell monitoring data matrix; Comparing the cell voltage in the cell monitoring data matrix with a preset voltage range to obtain a voltage over-limit mark, comparing the temperature difference between adjacent cells with a preset temperature difference threshold to obtain a temperature difference over-limit mark, and comparing the balanced current with a current threshold of the rated current to obtain a current over-limit mark; Generate an over-limit level matrix according to the voltage over-limit mark, the temperature difference over-limit mark and the current over-limit mark, and prioritize the over-limit level matrix to obtain a protection action instruction; When the protection action instruction is to reduce the balancing current, the balancing current value is decreased and calculated to obtain a reduced balancing current value, and the reduced balancing current value is updated to the main balancing circuit or the secondary balancing circuit; When the protection action instruction is to interrupt the balanced charging, a balanced interrupt signal sequence is generated, and the balanced interrupt signal sequence is input into the switch tube control terminals of the main balanced circuit and the secondary balanced circuit respectively, so that the switch tube enters the cut-off state; Feedback detection is performed on the derating equalizing current value or the execution state of the equalizing interrupt signal sequence to obtain a protection execution result, and the protection execution result is recorded in the equalizing charging log.

8. A balanced charging device for a power battery pack, characterized in that: A method for performing balanced charging of a power battery pack according to any one of claims 1 to 7, comprising: The parameter acquisition module is used to collect parameters of the power battery pack, measure the terminal voltage value, temperature value and charge and discharge current value of each single battery, and calculate the state of charge value and internal resistance change rate of each single battery; A feature extraction module, used to construct the terminal voltage value, the temperature value, the charge and discharge current value, the state of charge value and the internal resistance change rate into a state feature sequence, and input it into a multi-head self-attention network for feature extraction to generate a state correlation feature matrix; A construction module, used to construct a three-layer fully connected Actor network and a Critic network with a dual-Q structure based on the state association feature matrix, and output a balanced current value, a constant current or constant voltage balanced mode and a balanced time through the Actor network; The control module is used to control the DC-DC converter in the main balancing loop to generate a constant current output, or control the Buck-Boost converter in the secondary balancing loop to generate a constant voltage output according to the balancing current value, the balancing mode and the balancing time.

9. A balanced charging device for a power battery pack, characterized in that: The equalization charging device of the power battery pack includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instruction in the memory to enable the balanced charging device of the power battery pack to execute the balanced charging method for the power battery pack according to any one of claims 1 to 7.

Images