Pure electric vehicle charging control method and system

Abstract

The present invention provides a pure electric vehicle charging control method and system, which relates to the field of electric vehicle energy storage technology, including collecting real-time charging data, building a battery status evaluation model in combination with historical data, and calculating battery status information in real time; establishing a multi-objective optimization model based on the battery status information, and using an improved particle swarm algorithm to solve and obtain the optimal charging strategy; establishing a multi-dimensional safety monitoring system, executing a corresponding safety plan when charging abnormalities are detected, and feeding back to the multi-objective optimization model for strategy correction.

Description

Technical Field

[0001] The present invention relates to electric vehicle energy storage technology, and in particular to a pure electric vehicle charging control method and system. Background Art

[0002] Traditional charging control methods often use fixed charging strategies and cannot be dynamically adjusted according to the real-time status of the battery. This method ignores the performance changes of the battery in different use stages and environmental conditions, making it difficult to achieve the optimal charging effect, which may lead to low charging efficiency or shortened battery life.

[0003] Existing charging control systems lack a comprehensive assessment of the battery status. Most systems rely only on simple voltage and current monitoring and are unable to accurately estimate key parameters such as the battery’s remaining capacity, health status, and internal resistance. This lack of information can lead to incorrect decisions during charging, affecting charging performance and battery safety.

[0004] The current charging safety monitoring mechanism is relatively simple, mainly relying on a single threshold judgment. This method is difficult to deal with complex charging abnormalities and cannot identify and deal with potential safety risks in a timely manner. At the same time, the lack of an effective feedback mechanism makes it difficult to optimize and adjust the charging strategy in real time based on the execution results of the safety plan. Summary of the invention

[0005] The embodiments of the present invention provide a pure electric vehicle charging control method and system, which can solve the problems in the prior art.

[0006] According to a first aspect of the embodiments of the present invention,

[0007] Provides a pure electric vehicle charging control method, including:

[0008] Collecting real-time charging data during the charging process of the electric vehicle battery, the real-time charging data includes charging current, charging voltage, and battery temperature, building a battery status evaluation model in combination with historical charging data, and performing real-time calculations on the battery remaining capacity, health status, and internal resistance estimation value according to the battery status evaluation model to obtain the current battery status information;

[0009] Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established, and an improved particle swarm algorithm is used to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; and charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state evaluation model;

[0010] Based on the current status information of the battery and the real-time charging data during the charging process, a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring is established. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering the current limiting mode, starting forced voltage reduction and adjusting the heat dissipation power of the active heat dissipation device, and the execution result of the safety plan is fed back to the multi-objective optimization model to correct the optimal charging strategy.

[0011] In an optional embodiment,

[0012] The steps of collecting real-time charging data during the charging process of the electric vehicle battery, wherein the real-time charging data includes charging current, charging voltage, and battery temperature, building a battery status evaluation model in combination with historical charging data, and performing real-time calculation of the battery remaining capacity, health status, and internal resistance estimation value according to the battery status evaluation model to obtain the current battery status information include:

[0013] Collect charging current, charging voltage and battery temperature data during the charging process of the electric vehicle battery, use wavelet transform to perform noise reduction on the charging current, charging voltage and battery temperature data, select db4 wavelet basis for signal decomposition, use soft threshold method to process high frequency coefficients and reconstruct to obtain noise reduction signal;

[0014] The noise reduction signal is stored to form historical charging data, the historical charging data is segmented according to the charging cycle, the charging curve slope change characteristics, the voltage curve inflection point characteristics, the temperature change trend characteristics and the charging and discharging efficiency characteristics are extracted and normalized, a feature vector matrix is ​​constructed, and a principal component analysis method is used to reduce the dimension of the feature vector matrix to obtain historical feature data;

[0015] Based on the noise reduction signal and the historical feature data, a battery status assessment model including a bidirectional long short-term memory network layer and a self-attention mechanism is constructed, the weights of data at different time steps are dynamically adjusted through the self-attention mechanism, and the battery status assessment model is used to output the remaining capacity, health status and internal resistance estimation of the battery to obtain an assessment result of the current status information of the battery;

[0016] A confidence interval is calculated for the evaluation result. When it is detected that the confidence interval exceeds a preset threshold range, a data sampling strategy is adjusted and parameters of the battery status evaluation model are updated, and the evaluation result and its confidence interval are output as the current battery status information.

[0017] In an optional embodiment,

[0018] Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established, and an improved particle swarm algorithm is used to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; and the charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state evaluation model, comprising:

[0019] Obtaining a remaining capacity estimate, a health state estimate, and an internal resistance estimate of the battery, and establishing a dynamic constraint model for the charging process, wherein the dynamic constraint model includes a charging cutoff voltage constraint based on the remaining capacity estimate and the health state estimate, a maximum charging current constraint based on the health state estimate, and a charging power constraint based on the internal resistance estimate;

[0020] A multi-objective optimization model is constructed based on the charging cut-off voltage constraint, the maximum charging current constraint and the charging power constraint, wherein the multi-objective optimization model includes a charging time objective function, a charging efficiency objective function and a battery life objective function, wherein the charging time objective function is constructed based on the charging stage duration and the charging stage switching time, the charging efficiency objective function is constructed based on the internal resistance loss and the actual stored energy, and the battery life objective function is constructed based on the battery health state weighting function and the charging stress function;

[0021] The current value and duration of the charging phase are encoded as particle position vectors, adaptive inertia weights are designed based on health state estimates, local search terms based on charging efficiency are introduced into the particle movement speed update equation, and a penalty function for violating constraints is designed to form an improved particle swarm algorithm.

[0022] The improved particle swarm algorithm is used to solve the multi-objective optimization model, and the charging process is divided into a pre-charging stage, a constant current charging stage, and a constant voltage charging stage according to the solution result, and the charging current value and charging duration of each stage are determined to obtain the optimal charging strategy;

[0023] The charging data during the execution of the optimal charging strategy is collected, a plurality of charging data are selected as new training samples using a sliding time window, and based on the new training samples, the parameters of the battery state evaluation model are updated using an incremental learning method.

[0024] In an optional embodiment,

[0025] The steps of constructing the battery life objective function based on the battery health state weighting function and the charging stress function include:

[0026] Constructing a battery health state weighting function, wherein the battery health state weighting function is inversely proportional to the health state estimation value;

[0027] Constructing a charging stress function, wherein the charging stress function includes a current stress term, a temperature stress term, and a time stress term, wherein the current stress term is proportional to the square value of the charging current, the temperature stress term is proportional to the exponential function of the battery surface temperature, and the time stress term is proportional to the charging duration;

[0028] The product of the weighted function of the battery health state and the charging stress function during the entire charging process is time-integrated to obtain the battery life objective function.

[0029] In an optional embodiment,

[0030] The steps of designing an adaptive inertia weight based on the estimated health state, introducing a local search term based on charging efficiency into the particle movement speed update equation, and designing a penalty function for violating the constraint include:

[0031] The adaptive inertia weight is determined by a nonlinear mapping relationship between the health state estimation value and the preset upper and lower limits of the inertia weight, so that the adaptive inertia weight increases when the battery health state is lower than the preset battery health state threshold, and decreases when the battery health state is higher than the preset battery health state threshold;

[0032] The adaptive inertia weight is used to update the particle movement speed, and a local search term based on charging efficiency is introduced into the particle movement speed update equation, wherein the local search term is determined by the difference between the charging efficiency corresponding to the current particle charging strategy and the average charging efficiency of the population and the local optimal solution;

[0033] A penalty function is constructed based on voltage constraints, current constraints and power constraints in the charging process, wherein the penalty function includes a voltage over-limit penalty term, a current over-limit penalty term and a power over-limit penalty term, and the penalty factor in the penalty function is obtained by multiplying an initial penalty factor by a constraint violation adjustment term, wherein the constraint violation adjustment term is determined by multiplying a constraint adjustment coefficient by a constraint violation amount, and the constraint violation amount is determined by a difference between an actual value and a constraint upper limit value determined based on a health state estimate and an internal resistance estimate.

[0034] In an optional embodiment,

[0035] According to the current state information of the battery and the real-time charging data during the charging process, a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring is established. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting a forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and feeding back the execution result of the safety plan to the multi-objective optimization model for the optimal charging strategy correction step, including:

[0036] Collecting the current state information of the battery and the real-time charging data during the charging process, calculating the current difference between adjacent sampling periods to obtain the current mutation amount, comparing the current mutation amount with a preset current mutation threshold, and when the current mutation amount exceeds the preset current mutation threshold, determining a current limiting coefficient based on the rated charging current, limiting the charging current output by the multi-objective optimization model by the current limiting coefficient, and obtaining the actual charging current in the current limiting mode;

[0037] The theoretical charging voltage is calculated according to the open circuit voltage, internal resistance and polarization characteristics of the battery, and the deviation rate between the theoretical charging voltage and the measured charging voltage is compared with a preset voltage deviation threshold. When the deviation rate exceeds the preset voltage deviation threshold, a safety margin voltage is set based on the theoretical charging voltage, the charging voltage is forced to be reduced by the safety margin voltage, and the voltage upper limit after the voltage reduction is updated to the multi-objective optimization model;

[0038] Monitor the time gradient and spatial gradient of the battery temperature, where the time gradient is calculated by the temperature change rate of adjacent sampling periods, and the spatial gradient is calculated by the maximum temperature difference of each temperature measuring point in the battery pack. When the time gradient or the spatial gradient exceeds the corresponding preset temperature threshold, the heat dissipation power of the active heat dissipation device is determined based on the deviation between the real-time temperature and the target temperature and the temperature change rate, and the charging power constraint in the multi-objective optimization model is adjusted according to the heat dissipation power.

[0039] In an optional embodiment,

[0040] The steps to determine the current limit factor based on the rated charging current include:

[0041] Based on the rated charging current of the battery, the current limiting control interval is divided according to the magnitude of the current mutation, and the current limiting control interval includes: a first control interval, a second control interval and a third control interval; the preset current mutation threshold is divided into a first preset threshold, a second preset threshold and a third preset threshold; the current mutation amount in the first control interval is greater than the first preset threshold and less than or equal to the second preset threshold, the current mutation amount in the second control interval is greater than the second preset threshold and less than or equal to the third preset threshold, and the current mutation amount in the third control interval is greater than the third preset threshold;

[0042] The reference value of the current limiting coefficient is set to 1 based on the rated charging current, and the current limiting coefficient is dynamically adjusted using a piecewise function for different current limiting control intervals:

[0043] When the current mutation amount is in the first control interval, the value range of the current limiting coefficient is between 0.8 and 1.0, and the specific value is calculated as follows: current limiting coefficient = 1-0.2×(current mutation amount-first preset threshold) / (second preset threshold-first preset threshold);

[0044] When the current mutation amount is in the second control interval, the value range of the current limiting coefficient is between 0.5 and 0.8, and the specific value is calculated as follows: current limiting coefficient = 0.8-0.3×(current mutation amount-second preset threshold) / (third preset threshold-second preset threshold);

[0045] When the current mutation amount is in the third control interval, the value range of the current limiting coefficient is between 0.2 and 0.5, and the specific value is calculated as follows: current limiting coefficient = 0.5-0.3×(current mutation amount-third preset threshold) / third preset threshold;

[0046] When it is detected that the current mutation amount is less than the first preset threshold value for multiple consecutive sampling cycles, the current limiting coefficient is increased by 0.1 in each sampling cycle until the current limiting coefficient is restored to the reference value.

[0047] According to a second aspect of the embodiments of the present invention,

[0048] Provide pure electric vehicle charging control system, including:

[0049] The first unit is used to collect real-time charging data during the charging process of the electric vehicle battery, wherein the real-time charging data includes charging current, charging voltage, and battery temperature, and to build a battery status evaluation model in combination with historical charging data. The battery remaining capacity, health status, and internal resistance estimation value are calculated in real time according to the battery status evaluation model to obtain the current battery status information;

[0050] The second unit is used to establish a multi-objective optimization model including charging time, charging efficiency and battery life based on the current state information of the battery, and adopt an improved particle swarm algorithm to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; and use the charging data after executing the optimal charging strategy as a new training sample to update the parameters of the battery state evaluation model;

[0051] The third unit is used to establish a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring according to the current state information of the battery and the real-time charging data during the charging process. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and the execution result of the safety plan is fed back to the multi-objective optimization model to correct the optimal charging strategy.

[0052] According to a third aspect of the embodiments of the present invention,

[0053] An electronic device is provided, comprising:

[0054] processor;

[0055] a memory for storing processor-executable instructions;

[0056] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0057] A fourth aspect of the embodiments of the present invention is:

[0058] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0059] The present invention collects real-time charging data and combines historical data to build a battery status assessment model, which can accurately estimate the remaining capacity, health status and internal resistance of the battery. Based on this information, a multi-objective optimization model is established and solved using an improved particle swarm algorithm to obtain the optimal charging strategy that balances charging time, charging efficiency and battery life. This method can significantly improve charging efficiency while extending the service life of the battery.

[0060] The present invention uses the charging data after executing the optimal charging strategy as a new training sample to update the parameters of the battery state assessment model, so that the model can continuously learn and adapt to the actual state changes of the battery. This adaptive optimization mechanism ensures that the charging strategy can be continuously optimized as the battery state changes, always maintaining the best performance.

[0061] The present invention establishes a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring, which can comprehensively monitor various key parameters in the charging process. When a charging anomaly is detected, the system can execute corresponding safety plans according to the type of anomaly, such as entering current limiting mode, starting forced voltage reduction or adjusting heat dissipation power. This multi-level safety protection mechanism greatly improves the safety of the charging process and effectively reduces the risk of safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of a pure electric vehicle charging control method according to an embodiment of the present invention;

[0063] Figure 2 It is a schematic diagram of the structure of a pure electric vehicle charging control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0065] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0066] Figure 1 FIG. 1 is a flow chart of a pure electric vehicle charging control method according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0067] Collecting real-time charging data during the charging process of the electric vehicle battery, the real-time charging data includes charging current, charging voltage, and battery temperature, building a battery status evaluation model in combination with historical charging data, and performing real-time calculations on the battery remaining capacity, health status, and internal resistance estimation value according to the battery status evaluation model to obtain the current battery status information;

[0068] Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established, and an improved particle swarm algorithm is used to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; and charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state evaluation model;

[0069] Based on the current status information of the battery and the real-time charging data during the charging process, a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring is established. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering the current limiting mode, starting forced voltage reduction and adjusting the heat dissipation power of the active heat dissipation device, and the execution result of the safety plan is fed back to the multi-objective optimization model to correct the optimal charging strategy.

[0070] For example, first, collect real-time charging data during the charging process of the electric vehicle battery. These data include charging current, charging voltage and battery temperature. The collection method can be achieved by setting a current sensor, a voltage sensor and a temperature sensor in the battery management system. For example, collect data every 1 second, the current accuracy is 0.1A, the voltage accuracy is 0.01V, and the temperature accuracy is 0.1℃.

[0071] Next, a battery status assessment model is constructed based on historical charging data. The historical charging data includes charging records in the past 30 days, and each record contains information such as charging start time, end time, and charging amount. Machine learning algorithms such as support vector machines (SVM) are used to train the model. Input features include charging current, voltage, temperature and other parameters, and the output is the remaining capacity, health status and internal resistance estimation of the battery.

[0072] Then, the constructed battery status assessment model is used to calculate the remaining capacity, health status and internal resistance estimation of the battery in real time to obtain the current battery status information. For example, a calculation is performed every 5 seconds to obtain the remaining capacity of the battery as 80%, the health status as 95%, and the internal resistance estimation as 100mΩ.

[0073] Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established. The charging time objective is to minimize the time required for charging, the charging efficiency objective is to maximize the charging efficiency, and the battery life objective is to minimize the impact of charging on the battery life. The improved particle swarm algorithm is used to solve the multi-objective optimization model and obtain the optimal charging strategy.

[0074] The optimal charging strategy includes the charging current value and charging duration in different charging stages. For example, the first stage uses 2C (1C means the current that fully charges the rated capacity of the battery in 1 hour) constant current charging for 15 minutes, the second stage uses 1C constant current charging for 30 minutes, and the third stage uses 0.5C constant current charging until it is fully charged.

[0075] The charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state assessment model. Online learning algorithms, such as stochastic gradient descent, can be used to update the model parameters in real time to adapt to the dynamic changes in battery performance.

[0076] At the same time, a multi-dimensional safety monitoring system is established based on the current status information of the battery and the real-time charging data during the charging process. The system includes three dimensions: current mutation detection, voltage anomaly detection and temperature gradient monitoring.

[0077] Current mutation detection: Set the current change threshold to 0.5C / s. When the current change rate exceeds this threshold, it is determined as a current mutation.

[0078] Abnormal voltage detection: Set the normal working voltage range according to the battery type, for example, for ternary lithium batteries, set it to 2.5V-4.2V. When the battery voltage is detected to be outside this range, it is judged as abnormal voltage.

[0079] Temperature gradient monitoring: Set the temperature change threshold to 2°C / min. When the temperature change rate exceeds this threshold, it is judged as temperature abnormality.

[0080] When a charging anomaly is detected, the charging control system executes the corresponding safety plan according to the type of anomaly:

[0081] For current mutations, the system enters current limiting mode, reduces the charging current to 0.5C, and gradually returns to the original charging strategy within 30 seconds.

[0082] For voltage abnormalities, forced voltage reduction is initiated to reduce the charging voltage by 0.1V, and charging is suspended for 5 minutes before re-evaluating the battery status.

[0083] For abnormal temperatures, adjust the heat dissipation power of the active cooling device and increase the cooling fan speed by 50% for 10 minutes before re-evaluating the temperature condition.

[0084] The execution results of the safety plan will be fed back to the multi-objective optimization model to modify the optimal charging strategy. For example, if a temperature abnormality occurs, a temperature constraint can be added to the optimization model to limit the upper limit of the charging current to reduce heat generation.

[0085] The present invention improves charging efficiency and safety: by real-time monitoring of battery status and optimizing charging strategies, the charging speed is maximized and the charging time is shortened while ensuring charging safety; the battery life is extended: considering the dynamic changes in the battery health status, an adaptive charging strategy is adopted to reduce battery damage and extend the battery service life.

[0086] In an optional embodiment,

[0087] The steps of collecting real-time charging data during the charging process of the electric vehicle battery, wherein the real-time charging data includes charging current, charging voltage, and battery temperature, building a battery status evaluation model in combination with historical charging data, and performing real-time calculation of the battery remaining capacity, health status, and internal resistance estimation value according to the battery status evaluation model to obtain the current battery status information include:

[0088] Collect charging current, charging voltage and battery temperature data during the charging process of the electric vehicle battery, use wavelet transform to perform noise reduction on the charging current, charging voltage and battery temperature data, select db4 wavelet basis for signal decomposition, use soft threshold method to process high frequency coefficients and reconstruct to obtain noise reduction signal;

[0089] The noise reduction signal is stored to form historical charging data, the historical charging data is segmented according to the charging cycle, the charging curve slope change characteristics, the voltage curve inflection point characteristics, the temperature change trend characteristics and the charging and discharging efficiency characteristics are extracted and normalized, a feature vector matrix is ​​constructed, and a principal component analysis method is used to reduce the dimension of the feature vector matrix to obtain historical feature data;

[0090] Based on the noise reduction signal and the historical feature data, a battery status assessment model including a bidirectional long short-term memory network layer and a self-attention mechanism is constructed, the weights of data at different time steps are dynamically adjusted through the self-attention mechanism, and the battery status assessment model is used to output the remaining capacity, health status and internal resistance estimation of the battery to obtain an assessment result of the current status information of the battery;

[0091] A confidence interval is calculated for the evaluation result. When it is detected that the confidence interval exceeds a preset threshold range, a data sampling strategy is adjusted and parameters of the battery status evaluation model are updated, and the evaluation result and its confidence interval are output as the current battery status information.

[0092] For example, real-time charging data of electric vehicle batteries during charging is collected, including charging current, charging voltage and battery temperature. High-precision data acquisition equipment is used to record these parameters at intervals of 1 second. For example, some of the raw data collected during a charging process are: charging current 20.5A, charging voltage 375.2V, battery temperature 28.3℃.

[0093] The collected raw data is subjected to wavelet transform denoising. The db4 wavelet basis is selected to perform a 4-layer decomposition of the signal. Taking the charging current data as an example, it is decomposed into approximate components and detail components. The high-frequency coefficients are processed using the soft threshold method, and the threshold is set to 3σ (σ is the standard deviation of the noise). After processing, the signal is reconstructed to obtain the denoised charging current data, and the noise is effectively suppressed.

[0094] The denoised data is stored in segments according to the charging cycle to form a historical charging database. Each charging cycle contains a complete charging and discharging process. Features are extracted from the historical data, including the charging curve slope change feature, the voltage curve inflection point feature, the temperature change trend feature, and the charging and discharging efficiency feature. Taking the charging curve slope change as an example, the rate of change of the charging current over time is calculated to obtain the slope sequence. All features are normalized to construct a feature vector matrix.

[0095] Perform principal component analysis on the feature vector matrix and retain the principal components with a cumulative contribution rate of 95%. For example, if the original feature dimension is 20, 8 principal components are retained after dimension reduction by principal component analysis to obtain the historical feature data after dimension reduction.

[0096] A battery status assessment model is constructed based on the noise reduction signal and historical feature data. The model contains a bidirectional long short-term memory network layer and a self-attention layer. The bidirectional long short-term memory network layer is used to capture time series features, and the self-attention layer is used to dynamically adjust the weights of data at different time steps. The model input is the current charging data and historical feature data, and the output is the battery remaining capacity, health status, and internal resistance estimation.

[0097] When training the model, the Adam optimizer is used, the learning rate is set to 0.001, the batch size is 64, and the number of training rounds is 100. The mean square error is used as the loss function. After training, the test set is used to evaluate the model performance to ensure the generalization ability of the model.

[0098] Use the trained model to evaluate the newly collected charging data in real time. For example, the evaluation result of one time is: the remaining battery capacity is 85%, the health status is 0.92, and the estimated internal resistance is 0.15Ω.

[0099] Calculate a 95% confidence interval for the evaluation results. If the confidence interval exceeds the preset threshold range (e.g., the width of the confidence interval for the remaining battery capacity is greater than 10%), adjust the data sampling strategy, such as increasing the sampling frequency or expanding the sampling range. At the same time, use the newly collected data to update the model parameters to improve the model accuracy.

[0100] The evaluation results and their confidence intervals are output as the current battery status information. For example, the output results are: battery remaining capacity 85±3%, health status 0.92±0.02, and internal resistance estimated value 0.15±0.01Ω.

[0101] The present invention adopts the principal component analysis method to reduce the dimension of the eigenvector matrix, reduces data redundancy and improves model training efficiency; combines the bidirectional long short-term memory network and the self-attention mechanism to construct a battery state assessment model that can effectively capture timing characteristics and key information, thereby improving the accuracy and reliability of the assessment results.

[0102] In an optional embodiment,

[0103] Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established, and an improved particle swarm algorithm is used to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; and the charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state evaluation model, comprising:

[0104] Obtaining a remaining capacity estimate, a health state estimate, and an internal resistance estimate of the battery, and establishing a dynamic constraint model for the charging process, wherein the dynamic constraint model includes a charging cutoff voltage constraint based on the remaining capacity estimate and the health state estimate, a maximum charging current constraint based on the health state estimate, and a charging power constraint based on the internal resistance estimate;

[0105] A multi-objective optimization model is constructed based on the charging cut-off voltage constraint, the maximum charging current constraint and the charging power constraint, wherein the multi-objective optimization model includes a charging time objective function, a charging efficiency objective function and a battery life objective function, wherein the charging time objective function is constructed based on the charging stage duration and the charging stage switching time, the charging efficiency objective function is constructed based on the internal resistance loss and the actual stored energy, and the battery life objective function is constructed based on the battery health state weighting function and the charging stress function;

[0106] The current value and duration of the charging phase are encoded as particle position vectors, adaptive inertia weights are designed based on health state estimates, local search terms based on charging efficiency are introduced into the particle movement speed update equation, and a penalty function for violating constraints is designed to form an improved particle swarm algorithm.

[0107] The improved particle swarm algorithm is used to solve the multi-objective optimization model, and the charging process is divided into a pre-charging stage, a constant current charging stage, and a constant voltage charging stage according to the solution result, and the charging current value and charging duration of each stage are determined to obtain the optimal charging strategy;

[0108] The charging data during the execution of the optimal charging strategy is collected, a plurality of charging data are selected as new training samples using a sliding time window, and based on the new training samples, the parameters of the battery state evaluation model are updated using an incremental learning method.

[0109] Exemplarily, the estimated value of the remaining capacity, the estimated value of the health state, and the estimated value of the internal resistance of the battery are obtained. For example, the battery management system collects parameters such as battery voltage, current, and temperature, and uses the Kalman filter algorithm to estimate that the remaining capacity is 80%, the health state is 90%, and the internal resistance is 50 mΩ.

[0110] Based on these estimated values, a dynamic constraint model of the charging process is established. The charging cut-off voltage constraint is set to 4.2V, the maximum charging current constraint is set to 1C rate, and the charging power constraint is ≤500W.

[0111] A multi-objective optimization model is constructed, including three objective functions: charging time, charging efficiency, and battery life. The charging time objective function considers the duration of each charging stage and the switching time. The charging efficiency objective function is calculated based on the internal resistance loss and the actual stored energy. The battery life objective function is constructed by weighting the health state and the charging stress function.

[0112] The current value and duration of the charging phase are encoded as the particle position vector. For example, the pre-charging phase current is 0.1C and lasts for 10 minutes, the constant current charging phase current is 0.5C and lasts for 60 minutes, and the constant voltage charging phase current is 0.05C and lasts for 30 minutes, which can be encoded as [0.1, 10, 0.5, 60, 0.05, 30].

[0113] The adaptive inertia weight is designed to be 0.7 based on the estimated value of 90% of the health state. The local search term based on the charging efficiency is introduced into the particle movement speed update equation, and the search coefficient is set to 0.2. The penalty function for violating the constraints is designed, and the penalty coefficient for exceeding the constraints is 100.

[0114] The improved particle swarm algorithm is used to solve the multi-objective optimization model. The population size is set to 50, and the maximum number of iterations is 100. The optimal solution is selected by non-dominated sorting and crowding calculation, and the optimal charging strategy is obtained: the current in the pre-charging stage is 0.05C and lasts for 15 minutes; the current in the constant current charging stage is 0.8C and lasts for 45 minutes; the current in the constant voltage charging stage is 0.1C and lasts for 40 minutes.

[0115] Execute the optimal charging strategy and collect data such as voltage, current, and temperature during the charging process. Use a 5-minute sliding time window and select multiple charging data as new training samples. For example, select data [4.0V, 0.05C, 25°C], [4.1V, 0.8C, 26°C], [4.2V, 0.1C, 27°C] at t=5min, 10min, and 15min.

[0116] Based on new training samples, the parameters of the battery state assessment model are updated using incremental learning. For example, the state transfer matrix and observation matrix of the Kalman filter model are updated online so that the model can adapt to the dynamic changes in battery performance.

[0117] The present invention achieves a comprehensive balance among charging time, efficiency and lifespan through multi-objective optimization, thereby improving the comprehensive performance of the charging strategy; adopts an improved particle swarm algorithm to improve the solution efficiency and accuracy, thereby ensuring the optimality of the charging strategy; and continuously optimizes the battery state assessment model through incremental learning, thereby improving the accuracy and adaptability of the model.

[0118] In an optional embodiment,

[0119] The steps of constructing the battery life objective function based on the battery health state weighting function and the charging stress function include:

[0120] Constructing a battery health state weighting function, wherein the battery health state weighting function is inversely proportional to the health state estimation value;

[0121] Constructing a charging stress function, wherein the charging stress function includes a current stress term, a temperature stress term, and a time stress term, wherein the current stress term is proportional to the square value of the charging current, the temperature stress term is proportional to the exponential function of the battery surface temperature, and the time stress term is proportional to the charging duration;

[0122] The product of the weighted function of the battery health state and the charging stress function during the entire charging process is time-integrated to obtain the battery life objective function.

[0123] Exemplarily, a method for constructing a battery life objective function based on battery health status and charging stress uses the product of a battery health status weighted function and a charging stress function to perform a time integration operation to achieve an accurate assessment of battery life loss during the charging process.

[0124] The process of constructing the battery health state weighting function first determines the health state weighting coefficient, which is adjusted according to different battery types and usage scenarios, and usually takes values ​​between 0.8 and 1.2. For general-purpose lithium-ion power batteries, the health state weighting coefficient can be 1. At this time, if the battery health state estimate is 0.9, the health state weighting function value is 1.11. The health state weighting function is inversely proportional to the health state estimate. As the battery health state decreases, the weighting function value increases accordingly, reflecting the fact that batteries with poor health are more sensitive to charging stress.

[0125] The construction of the charging stress function includes three parts: current stress term, temperature stress term and time stress term. The current stress term is determined by the product of the current stress coefficient and the square value of the charging current. The current stress coefficient is determined according to the battery capacity and the maximum allowable charging current, and the value range is between 0.001 and 0.01. The temperature stress term is expressed by the product of the temperature stress coefficient and the battery surface temperature to the power of the natural logarithm base, in which the reference temperature and temperature scaling factor need to be introduced. The temperature stress coefficient ranges from 0.01 to 0.1. The time stress term is determined by the product of the time stress coefficient and the charging duration, and the time stress coefficient ranges from 0.001 to 0.01.

[0126] Taking the power battery as an example, the current stress coefficient is 0.005, the temperature stress coefficient is 0.05, and the time stress coefficient is 0.003. When the charging current is 2 amperes, the battery surface temperature is 30 degrees Celsius, and the charging has been 30 minutes, the reference temperature is set to 25 degrees Celsius, and the temperature scaling factor is set to 10, the values ​​of each stress item can be calculated. The current stress item is 0.02, the temperature stress item is related to the value of the battery surface temperature exceeding the reference temperature, and the time stress item is 0.09.

[0127] In order to obtain the battery life objective function value, it is necessary to perform a time integration operation on the product of the battery health state weighted function and the charging stress function during the entire charging process. In specific implementation, the charging process is divided into time segments with an interval of 5 minutes. In each time segment, the product of the health state weighted function value and the charging stress function value is calculated and multiplied by the time interval. The calculation results of all time segments are summed to obtain the final value of the battery life objective function.

[0128] The present invention realizes differentiated evaluation of batteries in different health states by introducing a battery health state weighting function; the charging stress function comprehensively considers three key factors: current, temperature and time, and comprehensively reflects the impact of the charging process on the battery life; the time integration method is used to evaluate the entire charging process, which provides a reliable basis for optimizing the charging strategy and effectively guides the formulation and implementation of the charging control strategy.

[0129] In an optional embodiment,

[0130] The steps of designing an adaptive inertia weight based on the estimated health state, introducing a local search term based on charging efficiency into the particle movement speed update equation, and designing a penalty function for violating the constraint include:

[0131] The adaptive inertia weight is determined by a nonlinear mapping relationship between the health state estimation value and the preset upper and lower limits of the inertia weight, so that the adaptive inertia weight increases when the battery health state is lower than the preset battery health state threshold, and decreases when the battery health state is higher than the preset battery health state threshold;

[0132] The adaptive inertia weight is used to update the particle movement speed, and a local search term based on charging efficiency is introduced into the particle movement speed update equation, wherein the local search term is determined by the difference between the charging efficiency corresponding to the current particle charging strategy and the average charging efficiency of the population and the local optimal solution;

[0133] A penalty function is constructed based on voltage constraints, current constraints and power constraints in the charging process, wherein the penalty function includes a voltage over-limit penalty term, a current over-limit penalty term and a power over-limit penalty term, and the penalty factor in the penalty function is obtained by multiplying an initial penalty factor by a constraint violation adjustment term, wherein the constraint violation adjustment term is determined by multiplying a constraint adjustment coefficient by a constraint violation amount, and the constraint violation amount is determined by a difference between an actual value and a constraint upper limit value determined based on a health state estimate and an internal resistance estimate.

[0134] For example, first, obtain the estimated value of the current health state of the battery. The battery management system can collect the battery's voltage, current, temperature and other parameters in real time, and use the Kalman filter-based state estimation algorithm to obtain the estimated value of the battery health state in combination with the battery's historical usage data. For example, the estimated value of the health state of a battery calculated is 85%.

[0135] Then, an adaptive inertia weight is designed based on the obtained health state estimate. The upper limit of the inertia weight is set to 0.9, the lower limit is set to 0.4, and the health state threshold is set to 80%. When the health state estimate is lower than 80%, a nonlinear mapping is used to make the inertia weight closer to 0.9; when it is higher than 80%, the inertia weight is made closer to 0.4. For the case of a health state of 85%, the corresponding adaptive inertia weight is 0.65.

[0136] Next, the adaptive inertia weight and the local search term based on charging efficiency are introduced into the speed update equation of the particle swarm algorithm. The local search term is determined by the difference between the charging efficiency of the current particle charging strategy and the average charging efficiency of the population, as well as the local optimal solution. For example, the current particle charging efficiency is 92%, the average charging efficiency of the population is 90%, and the difference is 2%. The local search term is calculated in combination with the local optimal solution.

[0137] Then, a penalty function containing voltage, current, and power constraints is constructed. Taking the voltage constraint as an example, the initial penalty factor is set to 10 and the constraint adjustment coefficient is set to 0.5. Based on the 85% health state and the estimated internal resistance value, the voltage constraint upper limit is calculated to be 4.2V. If the actual voltage is 4.3V and the constraint violation is 0.1V, the constraint violation adjustment item is 0.5*0.1=0.05, and the final voltage overlimit penalty factor is 10*1.05=10.5. The penalty factors for current and power are calculated in the same way.

[0138] Next, the charging strategy is iteratively optimized using the updated particle velocity equation and penalty function. In each iteration, the particle position is first updated according to the particle velocity, that is, the charging strategy; then the objective function value under the new strategy is calculated, and the penalty function value is added as the particle fitness; finally, the individual optimal solution and the global optimal solution are updated.

[0139] Finally, when the number of iterations reaches the set value or there is no significant improvement after multiple consecutive optimal solutions, the optimal charging strategy is output as the optimization result. This strategy comprehensively considers the battery health status, charging efficiency and various constraints, and can achieve fast and safe battery charging.

[0140] This method improves the adaptability of the algorithm to batteries in different health states and enhances the robustness of charging strategy optimization by introducing adaptive inertia weights; the local search term based on charging efficiency helps the algorithm to quickly converge to an efficient charging strategy and improves the optimization efficiency; the designed penalty function can effectively avoid safety hazards such as overvoltage and overcurrent during the charging process, ensuring the safety of battery charging.

[0141] In an optional embodiment,

[0142] According to the current state information of the battery and the real-time charging data during the charging process, a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring is established. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting a forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and feeding back the execution result of the safety plan to the multi-objective optimization model for the optimal charging strategy correction step, including:

[0143] Collecting the current state information of the battery and the real-time charging data during the charging process, calculating the current difference between adjacent sampling periods to obtain the current mutation amount, comparing the current mutation amount with a preset current mutation threshold, and when the current mutation amount exceeds the preset current mutation threshold, determining a current limiting coefficient based on the rated charging current, limiting the charging current output by the multi-objective optimization model by the current limiting coefficient, and obtaining the actual charging current in the current limiting mode;

[0144] The theoretical charging voltage is calculated according to the open circuit voltage, internal resistance and polarization characteristics of the battery, and the deviation rate between the theoretical charging voltage and the measured charging voltage is compared with a preset voltage deviation threshold. When the deviation rate exceeds the preset voltage deviation threshold, a safety margin voltage is set based on the theoretical charging voltage, the charging voltage is forced to be reduced by the safety margin voltage, and the voltage upper limit after the voltage reduction is updated to the multi-objective optimization model;

[0145] Monitor the time gradient and spatial gradient of the battery temperature, where the time gradient is calculated by the temperature change rate of adjacent sampling periods, and the spatial gradient is calculated by the maximum temperature difference of each temperature measuring point in the battery pack. When the time gradient or the spatial gradient exceeds the corresponding preset temperature threshold, the heat dissipation power of the active heat dissipation device is determined based on the deviation between the real-time temperature and the target temperature and the temperature change rate, and the charging power constraint in the multi-objective optimization model is adjusted according to the heat dissipation power.

[0146] Exemplarily, the current state information of the battery and the real-time charging data during the charging process are first collected. The current state information of the battery includes but is not limited to the open circuit voltage, internal resistance, polarization characteristics, capacity and health status of the battery. Real-time charging data includes charging current, charging voltage, battery temperature, etc. These data are obtained through various sensors and measuring devices of the battery management system, and the sampling frequency can be set to once per second.

[0147] Next, based on the collected data, multi-dimensional safety monitoring is implemented. First, current mutation detection is performed. The current difference between adjacent sampling cycles is calculated to obtain the current mutation amount. For example, if the charging current in the current sampling cycle is 10A and the previous sampling cycle is 9A, the current mutation amount is 1A. This current mutation amount is compared with the preset current mutation threshold (such as 0.5A). When the current mutation amount exceeds the threshold, it is considered that an abnormal current mutation has occurred.

[0148] When a sudden change in current is detected, the system will start the current limiting mode. The specific implementation method is to determine a current limiting coefficient based on the rated charging current. For example, if the rated charging current is 20A and the current charging current is 15A, the current limiting coefficient can be set to 0.8. Then multiply this current limiting coefficient by the charging current originally output by the multi-objective optimization model to obtain the actual charging current. This can effectively suppress abnormal current fluctuations.

[0149] Voltage anomaly detection is the second dimension of safety monitoring. First, the theoretical charging voltage is calculated based on the known open circuit voltage, internal resistance and polarization characteristics of the battery. For example, if the open circuit voltage is 3.7V, the internal resistance is 10mΩ, the charging current is 10A, and the equivalent internal resistance after considering the polarization effect is 15mΩ, then the theoretical charging voltage should be 3.85V. Compare this theoretical value with the actual measured charging voltage and calculate the deviation rate. Assuming the measured voltage is 3.95V, the deviation rate is 2.6%. If this deviation rate exceeds the preset threshold (such as 2%), it is determined that a voltage anomaly has occurred.

[0150] When voltage anomalies are detected, the system will initiate forced voltage reduction measures. A safety margin is set based on the theoretical charging voltage, such as 0.05V. Then the new upper limit of the charging voltage becomes 3.9V. The system will force the charging voltage to be controlled below this new upper limit, and synchronize this updated voltage limit to the multi-objective optimization model for subsequent charging strategy optimization.

[0151] Temperature monitoring is the third safety dimension, which includes two aspects: time gradient and spatial gradient. The time gradient is obtained by calculating the temperature change rate of adjacent sampling periods. For example, if the current temperature is 30°C and the temperature 1 second ago was 29.8°C, the time gradient is 0.2°C / s. The spatial gradient is obtained by calculating the maximum temperature difference of each temperature measurement point in the battery pack. Assuming that the battery pack has 10 temperature measurement points, the highest temperature is 32°C, and the lowest temperature is 30°C, the spatial gradient is 2°C.

[0152] When the time gradient or spatial gradient exceeds the preset threshold (such as the time gradient threshold of 0.1℃ / s, the spatial gradient threshold of 1.5℃), the system will adjust the working state of the active heat dissipation device. The adjustment is based on the deviation between the real-time temperature and the target temperature, as well as the temperature change rate. For example, if the target temperature is 25℃, the current temperature is 30℃, and the temperature change rate is 0.2℃ / s, the heat dissipation power may need to be increased to 80% of the rated power. Based on the calculated heat dissipation power, the system will adjust the charging power constraint in the multi-objective optimization model accordingly to balance the requirements of charging speed and temperature control.

[0153] In the process of executing the above safety plan, the system will record the execution results of each operation in real time, including the actual charging current after current limiting, the actual charging voltage after voltage reduction, the adjusted heat dissipation power, etc. These data will be fed back to the multi-objective optimization model in a timely manner for dynamic optimization and adjustment of the charging strategy. For example, if it is found that the voltage reduction measures cause a significant decrease in charging efficiency, the model may adjust the charging current curve accordingly to maximize charging efficiency while ensuring safety.

[0154] By establishing a multi-dimensional safety monitoring system, the present invention can promptly detect abnormal situations during the charging process and take corresponding safety plans, effectively reducing safety risks such as battery overcharging, over-discharging, and overheating, and improving the safety of the entire charging process; by feeding back the execution results of the safety plan to the multi-objective optimization model, dynamic adjustment and optimization of the charging strategy is achieved; this closed-loop control mechanism can maximize the charging efficiency and shorten the charging time while ensuring safety.

[0155] In an optional embodiment,

[0156] The steps to determine the current limit factor based on the rated charging current include:

[0157] Based on the rated charging current of the battery, the current limiting control interval is divided according to the magnitude of the current mutation, and the current limiting control interval includes: a first control interval, a second control interval and a third control interval; the preset current mutation threshold is divided into a first preset threshold, a second preset threshold and a third preset threshold; the current mutation amount in the first control interval is greater than the first preset threshold and less than or equal to the second preset threshold, the current mutation amount in the second control interval is greater than the second preset threshold and less than or equal to the third preset threshold, and the current mutation amount in the third control interval is greater than the third preset threshold;

[0158] The reference value of the current limiting coefficient is set to 1 based on the rated charging current, and the current limiting coefficient is dynamically adjusted using a piecewise function for different current limiting control intervals:

[0159] When the current mutation amount is in the first control interval, the value range of the current limiting coefficient is between 0.8 and 1.0, and the specific value is calculated as follows: current limiting coefficient = 1-0.2×(current mutation amount-first preset threshold) / (second preset threshold-first preset threshold);

[0160] When the current mutation amount is in the second control interval, the value range of the current limiting coefficient is between 0.5 and 0.8, and the specific value is calculated as follows: current limiting coefficient = 0.8-0.3×(current mutation amount-second preset threshold) / (third preset threshold-second preset threshold);

[0161] When the current mutation amount is in the third control interval, the value range of the current limiting coefficient is between 0.2 and 0.5, and the specific value is calculated as follows: current limiting coefficient = 0.5-0.3×(current mutation amount-third preset threshold) / third preset threshold;

[0162] When it is detected that the current mutation amount is less than the first preset threshold value for multiple consecutive sampling cycles, the current limiting coefficient is increased by 0.1 in each sampling cycle until the current limiting coefficient is restored to the reference value.

[0163] For example, in a method for determining a current limiting factor based on a rated charging current, it is first necessary to obtain the rated charging current of the battery as a reference value. This reference value can be obtained from the specification parameters of the battery or through actual testing. For example, for a typical lithium-ion battery, its rated charging current may be 2A.

[0164] Next, the current limiting control interval is divided according to the size of the current mutation. Here, the current limiting control interval is divided into three: the first control interval, the second control interval and the third control interval. In order to divide these intervals, three preset current mutation thresholds need to be set: the first preset threshold, the second preset threshold and the third preset threshold. The specific values ​​of these thresholds can be determined according to the actual application scenario and battery characteristics. For example, the first preset threshold can be set to 0.5A, the second preset threshold can be set to 1A, and the third preset threshold can be set to 1.5A.

[0165] Based on these thresholds, three control intervals can be defined: the current mutation amount in the first control interval is greater than 0.5A and less than or equal to 1A; the current mutation amount in the second control interval is greater than 1A and less than or equal to 1.5A; and the current mutation amount in the third control interval is greater than 1.5A.

[0166] After determining the control range, it is necessary to set the reference value of the current limiting coefficient. Here, the reference value is set to 1, indicating that no current limiting is performed under normal circumstances. Then, for different current limiting control ranges, the current limiting coefficient is dynamically adjusted using a piecewise function.

[0167] When the current mutation is in the first control interval, that is, greater than 0.5A and less than or equal to 1A, the current limiting coefficient ranges from 0.8 to 1.0. The specific value is calculated in the following way: first calculate the difference between the current mutation and the first preset threshold, and then divide it by the difference between the second preset threshold and the first preset threshold to obtain a proportional value between 0 and 1. Multiply this proportional value by 0.2, and then subtract the result from 1 to obtain the final current limiting coefficient.

[0168] For example, assuming that the detected current mutation is 0.8A, the calculation process of the current limiting coefficient is: first calculate (0.8A-0.5A) / (1A-0.5A)=0.6, and then calculate 1-0.2×0.6=0.88. Therefore, in this case, the current limiting coefficient is 0.88.

[0169] When the current mutation is in the second control interval, that is, greater than 1A and less than or equal to 1.5A, the current limiting coefficient ranges from 0.5 to 0.8. The specific value calculation method is similar to the first control interval, but the starting value and the range of variation are different. First, calculate the difference between the current mutation and the second preset threshold, and then divide it by the difference between the third preset threshold and the second preset threshold to obtain a proportional value between 0 and 1. Multiply this proportional value by 0.3, and then subtract the result from 0.8 to get the final current limiting coefficient.

[0170] For example, assuming that the detected current mutation is 1.3A, the calculation process of the current limiting coefficient is: first calculate (1.3A-1A) / (1.5A-1A)=0.6, and then calculate 0.8-0.3×0.6=0.62. Therefore, in this case, the current limiting coefficient is 0.62.

[0171] When the current mutation is in the third control interval, that is, greater than 1.5A, the current limiting coefficient ranges from 0.2 to 0.5. The specific value is calculated in a similar way, but considering that this is the highest control interval, the calculation method is slightly different. First, calculate the difference between the current mutation and the third preset threshold, and then divide it by the third preset threshold itself to get a proportional value. Multiply this proportional value by 0.3, and then subtract the result from 0.5 to get the final current limiting coefficient.

[0172] For example, assuming that the detected current mutation is 2A, the calculation process of the current limiting coefficient is: first calculate (2A-1.5A) / 1.5A=0.33, and then calculate 0.5-0.3×0.33=0.401. Therefore, in this case, the current limiting coefficient is 0.401.

[0173] In practical applications, it is necessary to continuously monitor the changes in the current mutation amount. When it is detected that the current mutation amount is less than the first preset threshold value for multiple consecutive sampling cycles, it means that the current has stabilized, and the current limiting can be gradually relaxed. The specific method is to increase the current limiting coefficient by 0.1 in each sampling cycle until the current limiting coefficient returns to the reference value of 1.

[0174] For example, assuming that the current limiting coefficient is 0.7, and the current mutation detected in three consecutive sampling cycles is less than 0.5A, then the change process of the current limiting coefficient is: it increases to 0.8 in the first cycle, increases to 0.9 in the second cycle, and increases to 1.0 in the third cycle. At this time, it returns to the baseline value and no longer limits the current.

[0175] The present invention can achieve refined control of the charging current by dividing the charging current into multiple current limiting control intervals and dynamically adjusting the current limiting coefficient using a piecewise function. It can take corresponding current limiting measures according to the degree of current mutation, which can effectively protect the battery without excessively limiting the charging speed. By setting multiple preset thresholds and corresponding current limiting coefficient calculation methods, the thresholds and calculation parameters can be adjusted according to different types of batteries and charging scenarios to obtain the best charging effect. By real-time monitoring of the current mutation amount and gradually relaxing the current limiting when it is stable, this dynamic adjustment mechanism can make the charging process more intelligent and efficient, extend the battery life while also improving the user experience.

[0176] Figure 2 FIG. 1 is a schematic diagram of a charging control system for a pure electric vehicle according to an embodiment of the present invention. Figure 2 As shown, the system comprises:

[0177] The first unit is used to collect real-time charging data during the charging process of the electric vehicle battery, wherein the real-time charging data includes charging current, charging voltage, and battery temperature, and to build a battery status evaluation model in combination with historical charging data. The battery remaining capacity, health status, and internal resistance estimation value are calculated in real time according to the battery status evaluation model to obtain the current battery status information;

[0178] The second unit is used to establish a multi-objective optimization model including charging time, charging efficiency and battery life based on the current state information of the battery, and adopt an improved particle swarm algorithm to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; and use the charging data after executing the optimal charging strategy as a new training sample to update the parameters of the battery state evaluation model;

[0179] The third unit is used to establish a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring according to the current state information of the battery and the real-time charging data during the charging process. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and the execution result of the safety plan is fed back to the multi-objective optimization model to correct the optimal charging strategy.

[0180] According to a third aspect of the embodiments of the present invention,

[0181] An electronic device is provided, comprising:

[0182] processor;

[0183] a memory for storing processor-executable instructions;

[0184] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0185] A fourth aspect of the embodiments of the present invention is:

[0186] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0187] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0188] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A pure electric vehicle charging control method, characterized in that: include: Collecting real-time charging data during the charging process of the electric vehicle battery, the real-time charging data includes charging current, charging voltage, and battery temperature, building a battery status evaluation model in combination with historical charging data, and performing real-time calculations on the battery remaining capacity estimation value, health status estimation value, and internal resistance estimation value according to the battery status evaluation model to obtain the current battery status information; Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established, and an improved particle swarm algorithm is used to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; The charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state evaluation model; According to the current state information of the battery and the real-time charging data during the charging process, a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring is established. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting a forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and the execution result of the safety plan is fed back to the multi-objective optimization model to correct the optimal charging strategy; The steps of collecting real-time charging data during the charging process of the electric vehicle battery, wherein the real-time charging data includes charging current, charging voltage, and battery temperature, building a battery status evaluation model in combination with historical charging data, and performing real-time calculations on the estimated value of the remaining battery capacity, the estimated value of the health status, and the estimated value of the internal resistance according to the battery status evaluation model to obtain the current battery status information include: Collect charging current, charging voltage and battery temperature data during the charging process of the electric vehicle battery, use wavelet transform to perform noise reduction on the charging current, charging voltage and battery temperature data, select db4 wavelet basis for signal decomposition, use soft threshold method to process high frequency coefficients and reconstruct to obtain noise reduction signal; The noise reduction signal is stored to form historical charging data, the historical charging data is segmented according to the charging cycle, the charging curve slope change characteristics, the voltage curve inflection point characteristics, the temperature change trend characteristics and the charging and discharging efficiency characteristics are extracted and normalized, a feature vector matrix is ​​constructed, and a principal component analysis method is used to reduce the dimension of the feature vector matrix to obtain historical feature data; Based on the noise reduction signal and the historical feature data, a battery status assessment model including a bidirectional long short-term memory network layer and a self-attention mechanism is constructed, the weights of data at different time steps are dynamically adjusted through the self-attention mechanism, and the battery status assessment model is used to output a battery remaining capacity estimation value, a health status estimation value, and an internal resistance estimation value, so as to obtain an assessment result of the current battery status information; A confidence interval is calculated for the evaluation result. When it is detected that the confidence interval exceeds a preset threshold range, a data sampling strategy is adjusted and parameters of the battery status evaluation model are updated, and the evaluation result and its confidence interval are output as the current battery status information.

2. The method according to claim 1, characterized in that Based on the current state information of the battery, a multi-objective optimization model including charging time, charging efficiency and battery life is established, and an improved particle swarm algorithm is used to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; The steps of using the charging data after executing the optimal charging strategy as a new training sample to update the parameters of the battery state evaluation model include: Obtaining a remaining capacity estimate, a health state estimate, and an internal resistance estimate of the battery, and establishing a dynamic constraint model for the charging process, wherein the dynamic constraint model includes a charging cutoff voltage constraint based on the remaining capacity estimate and the health state estimate, a maximum charging current constraint based on the health state estimate, and a charging power constraint based on the internal resistance estimate; A multi-objective optimization model is constructed based on the charging cut-off voltage constraint, the maximum charging current constraint and the charging power constraint, wherein the multi-objective optimization model includes a charging time objective function, a charging efficiency objective function and a battery life objective function, wherein the charging time objective function is constructed based on the charging stage duration and the charging stage switching time, the charging efficiency objective function is constructed based on the internal resistance loss and the actual stored energy, and the battery life objective function is constructed based on the battery health state weighting function and the charging stress function; The current value and duration of the charging phase are encoded as particle position vectors, adaptive inertia weights are designed based on health state estimates, local search terms based on charging efficiency are introduced into the particle movement speed update equation, and a penalty function for violating constraints is designed to form an improved particle swarm algorithm. The improved particle swarm algorithm is used to solve the multi-objective optimization model, and the charging process is divided into a pre-charging stage, a constant current charging stage, and a constant voltage charging stage according to the solution result, and the charging current value and charging duration of each stage are determined to obtain the optimal charging strategy; The charging data during the execution of the optimal charging strategy is collected, a plurality of charging data are selected as new training samples using a sliding time window, and based on the new training samples, the parameters of the battery state evaluation model are updated using an incremental learning method.

3. The method according to claim 2, characterized in that The steps of constructing the battery life objective function based on the battery health state weighting function and the charging stress function include: Constructing a battery health state weighting function, wherein the battery health state weighting function is inversely proportional to the health state estimation value; Constructing a charging stress function, wherein the charging stress function includes a current stress term, a temperature stress term, and a time stress term, wherein the current stress term is proportional to the square value of the charging current, the temperature stress term is proportional to the exponential function of the battery surface temperature, and the time stress term is proportional to the charging duration; The product of the weighted function of the battery health state and the charging stress function during the entire charging process is time-integrated to obtain the battery life objective function.

4. The method according to claim 2, characterized in that: The steps of designing an adaptive inertia weight based on the estimated health state, introducing a local search term based on charging efficiency into the particle movement speed update equation, and designing a penalty function for violating the constraint include: The adaptive inertia weight is determined by a nonlinear mapping relationship between the health state estimation value and the preset upper and lower limits of the inertia weight, so that the adaptive inertia weight increases when the battery health state is lower than the preset battery health state threshold, and decreases when the battery health state is higher than the preset battery health state threshold; The adaptive inertia weight is used to update the particle movement speed, and a local search term based on charging efficiency is introduced into the particle movement speed update equation, wherein the local search term is determined by the difference between the charging efficiency corresponding to the current particle charging strategy and the average charging efficiency of the population and the local optimal solution; A penalty function is constructed based on voltage constraints, current constraints and power constraints in the charging process, wherein the penalty function includes a voltage over-limit penalty term, a current over-limit penalty term and a power over-limit penalty term, and the penalty factor in the penalty function is obtained by multiplying an initial penalty factor by a constraint violation adjustment term, wherein the constraint violation adjustment term is determined by multiplying a constraint adjustment coefficient by a constraint violation amount, and the constraint violation amount is determined by a difference between an actual value and a constraint upper limit value determined based on a health state estimate and an internal resistance estimate.

5. The method according to claim 1, characterized in that According to the current state information of the battery and the real-time charging data during the charging process, a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring is established. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting a forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and feeding back the execution result of the safety plan to the multi-objective optimization model for the optimal charging strategy correction step, including: Collecting the current state information of the battery and the real-time charging data during the charging process, calculating the current difference between adjacent sampling periods to obtain the current mutation amount, comparing the current mutation amount with a preset current mutation threshold, and when the current mutation amount exceeds the preset current mutation threshold, determining a current limiting coefficient based on the rated charging current, limiting the charging current output by the multi-objective optimization model by the current limiting coefficient, and obtaining the actual charging current in the current limiting mode; The theoretical charging voltage is calculated according to the open circuit voltage, internal resistance and polarization characteristics of the battery, and the deviation rate between the theoretical charging voltage and the measured charging voltage is compared with a preset voltage deviation threshold. When the deviation rate exceeds the preset voltage deviation threshold, a safety margin voltage is set based on the theoretical charging voltage, the charging voltage is forced to be reduced by the safety margin voltage, and the voltage upper limit after the voltage reduction is updated to the multi-objective optimization model; Monitor the time gradient and spatial gradient of the battery temperature, where the time gradient is calculated by the temperature change rate of adjacent sampling periods, and the spatial gradient is calculated by the maximum temperature difference of each temperature measuring point in the battery pack. When the time gradient or the spatial gradient exceeds the corresponding preset temperature threshold, the heat dissipation power of the active heat dissipation device is determined based on the deviation between the real-time temperature and the target temperature and the temperature change rate, and the charging power constraint in the multi-objective optimization model is adjusted according to the heat dissipation power.

6. The method according to claim 5, characterized in that The steps to determine the current limit factor based on the rated charging current include: Based on the rated charging current of the battery, the current limiting control interval is divided according to the magnitude of the current mutation, and the current limiting control interval includes: a first control interval, a second control interval and a third control interval; the preset current mutation threshold is divided into a first preset threshold, a second preset threshold and a third preset threshold; the current mutation amount in the first control interval is greater than the first preset threshold and less than or equal to the second preset threshold, the current mutation amount in the second control interval is greater than the second preset threshold and less than or equal to the third preset threshold, and the current mutation amount in the third control interval is greater than the third preset threshold; The reference value of the current limiting coefficient is set to 1 based on the rated charging current, and the current limiting coefficient is dynamically adjusted using a piecewise function for different current limiting control intervals: When the current mutation amount is in the first control interval, the value range of the current limiting coefficient is between 0.8 and 1.0, and the specific value is calculated as follows: current limiting coefficient = 1-0.2×(current mutation amount-first preset threshold) / (second preset threshold-first preset threshold); When the current mutation amount is in the second control interval, the value range of the current limiting coefficient is between 0.5 and 0.8, and the specific value is calculated as follows: current limiting coefficient = 0.8-0.3×(current mutation amount-second preset threshold) / (third preset threshold-second preset threshold); When the current mutation amount is in the third control interval, the value range of the current limiting coefficient is between 0.2 and 0.5, and the specific value is calculated as follows: current limiting coefficient = 0.5-0.3×(current mutation amount-third preset threshold) / third preset threshold; When it is detected that the current mutation amount is less than the first preset threshold value for multiple consecutive sampling cycles, the current limiting coefficient is increased by 0.1 in each sampling cycle until the current limiting coefficient is restored to the reference value.

7. A pure electric vehicle charging control system, used to implement the method described in any one of claims 1 to 6, characterized in that: include: The first unit is used to collect real-time charging data during the charging process of the electric vehicle battery, wherein the real-time charging data includes charging current, charging voltage, and battery temperature, and to build a battery status evaluation model in combination with historical charging data. According to the battery status evaluation model, the estimated value of the remaining capacity of the battery, the estimated value of the health status, and the estimated value of the internal resistance of the battery are calculated in real time to obtain the current status information of the battery; The second unit is used to establish a multi-objective optimization model including charging time, charging efficiency and battery life based on the current state information of the battery, and adopt an improved particle swarm algorithm to solve the multi-objective optimization model to obtain an optimal charging strategy, wherein the optimal charging strategy includes a charging current value and a charging duration in a charging stage; The charging data after executing the optimal charging strategy is used as a new training sample to update the parameters of the battery state evaluation model; The third unit is used to establish a multi-dimensional safety monitoring system including current mutation detection, voltage anomaly detection and temperature gradient monitoring according to the current state information of the battery and the real-time charging data during the charging process. When a charging anomaly is detected, the charging control system executes a corresponding safety plan according to the type of anomaly. The safety plan includes entering a current limiting mode, starting forced voltage reduction and adjusting the heat dissipation power of an active heat dissipation device, and the execution result of the safety plan is fed back to the multi-objective optimization model to correct the optimal charging strategy.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images