Full-life-cycle optimal fast charging control method for battery system

Abstract

The invention relates to the technical field of power battery management, regulation and control, in particular to a battery system full-life-cycle optimal fast charge control method which comprises a frequency-band-divided cooperative control framework of high-frequency fast charge optimization, intermediate-frequency lithium precipitation detection and low-frequency parameter updating. According to the full-life-cycle optimal fast charge control method for the battery system, the battery system model is constructed to consider the inconsistency of parameters among the single cells, so that the situation that individual cells with parameter deviations reach the boundary in the charging process is avoided, and behaviors such as lithium precipitation and thermal runaway are prevented from being carried out to carry out health state estimation and model parameter (capacity and internal resistance) updating after the battery is aged; the problem that an aged battery system does not adapt to an original charging strategy can be avoided in time, online multi-target lossless optimal fast charging is achieved in a charging period, meanwhile, threshold constraints of internal negative electrode potential and external temperature are considered, other key side reaction problems such as lithium precipitation are avoided, and the service life of the battery system is prolonged. And the current is dynamically optimized by combining a closed-loop observer and an optimizer.

Description

Technical Field

[0001] The present invention relates to the technical field of power battery management and regulation, and specifically to an optimal fast charging control method for a battery system over its entire life cycle. Background Art

[0002] Currently, lithium-ion batteries have been widely used in electric vehicles and electronic devices, especially in the power battery systems of electric vehicles. The application of fast charging technology in electric vehicle systems has made significant progress, but the battery system also faces problems such as high-power fast charging that will cause the internal temperature of the battery system to rise, stress to increase, and induce side reactions, affecting battery life and safety. In addition, the deposition behavior of lithium ions on the negative electrode surface during fast charging (such as dendrite growth) may cause safety hazards. At present, there is still a lack of optimal fast charging control strategy development that is practical-oriented, oriented to actual electric vehicles, and takes into account the entire life cycle.

[0003] Since the research on inconsistency between cells in the existing battery system mainly focuses on the differences in the capacity, power, energy and other characteristics of the cells, the fast charging strategy actually applied to electric vehicles is still at the level of offline fast charging MAP based on experience. The current fast charging strategy has shortcomings in the ability to model the inconsistency within the battery system and between cells and update the fast charging strategy throughout the life cycle: 1) Inconsistency between cells in a battery pack (such as capacity / internal resistance differences) will lead to uneven current distribution during fast charging, exacerbating local lithium deposition and aging. However, existing fast charging strategies are difficult to solve such problems from a system-level model. 2) Most of the existing battery fast charging strategies do not consider the update of fast charging strategies after battery aging, and the few updates are only implemented based on experience, which makes it difficult to judge the pros and cons of the updated charging strategy, which easily leads to problems such as increased battery system safety risks and accelerated system aging; 3) As the number of cycles increases, the internal resistance of the battery increases and the capacity decreases. The original charging curve may cause overcharging or lithium deposition, but there is a lack of online strategy update mechanism based on aging status, and the existing SOH estimation model is not accurate enough to support the prediction of the correlation between attenuation and fast charging; 4) The development method has obvious limitations. The strategy is highly dependent on laboratory offline testing and cannot adapt to dynamic changes in actual working conditions (such as low temperature environment, system power, etc.). This causes the charging protocol to be disconnected from the real-time status of the vehicle, making it difficult to balance efficiency and safety.

[0004] In order to solve the problems existing in the above-mentioned existing battery fast charging control technology, an optimal fast charging control method for the entire life cycle of the battery system is proposed. Summary of the invention

[0005] 1. Technical issues to be resolved In view of the shortcomings of the prior art, the present invention provides an optimal fast charging control method for the entire life cycle of a battery system. It has the advantages of dividing the fast charging control into three-layer closed-loop control of high-medium-low frequency from a time scale, maximizing the charging rate within the safety threshold of the battery system and delaying the attenuation of the battery system, solving the problem that the existing power battery fast charging strategy is difficult to achieve optimal fast charging control throughout the entire life cycle.

[0006] (II) Technical solution In order to achieve the above-mentioned purpose of dividing the fast charging control into high-medium-low frequency three-layer closed-loop control from the time scale, maximizing the charging rate within the safety threshold range of the battery system and delaying the attenuation of the battery system, the present invention provides the following technical solutions: an optimal fast charging control method for the whole life cycle of a battery system, including a frequency-band collaborative control architecture of high-frequency fast charging optimization, medium-frequency lithium precipitation detection, and low-frequency parameter update, and the specific steps are divided into: S1. Build an equivalent model based on the power battery system, accurately estimate the internal and external potential and temperature information of the battery, and achieve multi-objective optimal fast charging control within a fast charging cycle; S2. According to the optimal charging current, the power battery system performs lithium plating warning detection after the charge and discharge cycle; S3. The power battery system estimates the health status and updates the model parameters; S4. After the equivalent model is updated, optimal fast charging control of the battery system over its entire life cycle is achieved based on the obtained optimal charging current.

[0007] Preferably, the multi-objective optimal fast charging control of the power battery system in a fast charging cycle specifically includes: 1) Establish a polarization equivalent circuit model based on the 3P6S module composed of monomer equivalent models, set model parameters and consider the parameter distribution between monomers; 2) Setting the objective function and constraints for optimal fast charging in the power battery system; 3) Construct an online state observer based on the equivalent model to automatically correct the voltage and current information estimated by the model at the next moment according to the BMS voltage and current signals collected during the charging process of the power battery system and the voltage and current information estimated by the model; 4) Build an online current optimizer based on the charging strategy, obtain the battery estimated state data according to the preset objective function and constraints and the real-time input state observer to obtain the optimal charging current at the next moment; 5) The charger obtains the optimal current at the next moment according to the online current optimizer to charge the power battery system.

[0008] Preferably, the specific steps of constructing the polarization equivalent circuit model include: 1) Set the polarization Rint model of thermoelectric coupling as the battery cell model, where the polarization Rint model includes the electrical model module and thermal model module of the battery cell. The model circuit principle is expressed as follows: , , in is the positive terminal voltage of the battery, is the negative terminal voltage of the battery, is the full battery terminal voltage, is the positive open circuit voltage of the battery, is the open circuit voltage of the negative electrode of the battery, is the full battery open circuit voltage, is the internal resistance of the battery positive electrode, is the internal resistance of the negative electrode of the battery, The current passing through the battery is set to the discharge current. Is a positive value, the current during charging is a negative value; 2) Based on the heat generation and heat dissipation principle of the lumped thermal model, a thermoelectric coupling model of the battery cell is constructed, which is expressed as: , in is the heat generated by the battery per unit time, is the battery temperature, is the heat transfer per unit time, is the heat transfer coefficient, is the battery surface area, is the ambient temperature, then the battery temperature The expression of time variation is: , in and Respectively represent the battery mass and battery specific heat capacity, and an equivalent model is established based on the thermoelectric coupling model; 3) Analyze the inconsistency of parameters between battery cells, including capacity , internal resistance and initial charge Inconsistency, set each parameter to conform to the normal distribution , then it is expressed as: , in is the initial capacity of the battery, is the internal resistance change rate of battery model B, and its mean 1, initial charge Corresponds to the initial SOC distribution and varies with the initial state of the system. The larger it is, the greater the deviation of the parameter distribution.

[0009] Preferably, the objective function represents minimizing the time for the power battery system to reach the target power, and the constraint conditions represent the maximum and minimum limits on the battery temperature, voltage, and charging current, specifically including: 1) When the time required for the power battery system to reach the target state of charge is set to be the shortest within a charging cycle, the objective function is defined as: , in Indicates the time Battery of the moment , The target power , set goals is 90%, Indicates the time The optimal current of the battery system at the moment; 2) Define the constraints as , The maximum charging current based on the 4C charging rate 232A, the maximum voltage of the battery cell 4.4V, the lowest voltage The lowest negative electrode potential threshold of the battery is 2.8V. 0.01V, the maximum battery temperature The lowest temperature is 50℃ -20℃, Estimate the negative electrode potential inside the battery at time k for the battery system.

[0010] Preferably, an online state observer based on an equivalent model is constructed based on an extended Kalman filter, and the terminal voltage and temperature at time k estimated by the battery system model are compared with the terminal voltage and temperature at time k collected by the actual BMS, and the terminal voltage, negative electrode potential and temperature estimation at time k+1 are corrected through gain; The online current optimizer is optimized according to the multi-particle swarm algorithm, and the output is obtained as The current at the moment is the optimal current The charger uses the optimal current at the next moment obtained by the optimizer to realize the charging process in the power battery system.

[0011] Preferably, the lithium deposition early warning method is a relaxation voltage differential method, which specifically includes: 1) Set the data collection duration to 600s~1200s, and select any time period within this duration to collect the static voltage and time data of the battery after fast charging; 2) Perform time differential analysis on each collected single-cell voltage signal to determine whether there is a minimum value. The differential analysis expression is: ,in is the sampling frequency, is the voltage difference, Count the sampling points; when When there is no minimum value, it is judged that the battery system has no lithium deposition or has a slight lithium deposition process within the acceptable threshold, and the multi-objective optimal fast charging control operation is repeated for the next charging; when When it is a minimum value, it is judged that there is a lithium deposition side reaction in the battery system, and the health status of the battery system is estimated and the model parameters are updated, including: 1) Estimate the capacity of the battery system and update the parameters based on the gated neural network; 2) When the battery system is charged for the next moment, the internal resistance of the battery at the current temperature and power is calculated based on the initial charging current. The calculation formula is: , in is the initial current, is the battery voltage mutation difference caused by charging for 10 seconds, is the total internal resistance of the battery during 10s of charging, It is the total internal resistance of the battery under the same SOC and temperature conditions within 10 seconds of charging before aging. is the internal resistance change value; According to the change of internal resistance Update the battery internal resistance, specifically: , in and are the internal resistance growth factors of the positive and negative electrodes, respectively, and , then set =0 and =1, thereby ensuring to the greatest extent that the updated fast charging strategy will not cause lithium plating at the negative electrode; 3) Based on the updated battery capacity and internal resistance parameter information, update the capacity, average value of positive electrode internal resistance and negative electrode internal resistance in the battery system model, and increase the parameter inconsistency between battery cells. The specific updates include: , according to the updated battery system model, charging is carried out in the next fast charging cycle based on the multi-objective optimal fast charging control method.

[0012] (III) Beneficial effects Compared with the prior art, the present invention provides an optimal fast charging control method for a battery system throughout its life cycle, which has the following beneficial effects: 1. The optimal fast charging control method for the battery system throughout its life cycle takes into account the inconsistency of parameters between cells by constructing a battery system model, avoiding the cells with individual parameter deviations from reaching their boundaries during charging, and preventing lithium deposition, thermal runaway and other behaviors, thereby optimizing the problem of ignoring cell differences in traditional strategies, resulting in uneven fast charging current distribution and accelerated local lithium deposition / aging.

[0013] 2. The optimal fast charging control method for the battery system throughout its life cycle can immediately avoid the problem of the aged battery system being unsuitable for the original charging strategy by estimating the health status and updating the model parameters (capacity and internal resistance) after battery aging, thereby optimizing the traditional strategy that cannot adapt to battery attenuation, and the problem of reduced charging speed or breach of safety thresholds after battery aging.

[0014] 3. The optimal fast charging control method for the battery system throughout its life cycle achieves online multi-objective lossless optimal fast charging within one charging cycle, while considering the threshold constraints of the internal negative electrode potential and the external temperature to avoid other key side reaction problems such as lithium plating. It also combines a closed-loop observer and optimizer to dynamically optimize the current, which can optimize the problems that only consider the influence of unilateral factors in traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the framework of the optimal fast charging control method for the entire life cycle of the power battery system of the present invention; Figure 2 It is a schematic diagram of the equivalent model structure of the battery system of the present invention; Figure 3 This is a schematic diagram of the optimization of the optimal fast charging current, power and voltage curves for an electric vehicle of the present invention; Figure 4 It is a schematic diagram of the battery system equivalent model parameter updating process of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the embodiments of the present invention and the accompanying drawings to clearly and completely describe the technical solutions 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.

[0017] Embodiment 1 An online state observer based on an equivalent model is constructed based on the extended Kalman filter. The terminal voltage and temperature at time k estimated by the battery system model are compared with the terminal voltage and temperature at time k collected by the actual BMS, and the estimated terminal voltage, negative electrode potential and temperature at time k+1 are corrected through gain.

[0018] Specifically, in this embodiment, an online state observer based on an equivalent model constructed based on an extended Kalman filter specifically includes: 1) Discretize the model and express it as: , in is the state transfer function, is the observation function, is the process noise, is the observation noise; 2) Initialize the state estimation and error covariance, respectively: State estimation initialization: , Error covariance initialization: ; 3) The initialization state estimation and error covariance prediction are: , , in is the Jacobian matrix; 4) Calculate the Kalman gain: , , According to the Kalman gain, the state estimation and covariance correction are expressed as: , , The terminal voltage and temperature at time k are estimated through the model and compared with the terminal voltage and temperature at time k collected by the actual BMS, and the terminal voltage, cathode potential and temperature estimation at time k+1 are corrected and optimized according to the Kalman gain.

[0019] Embodiment 2 Furthermore, the online current optimizer is optimized according to the multi-particle swarm algorithm, and the output is obtained as The current at the moment is the optimal current The charger uses the optimal current at the next moment obtained by the optimizer to realize the charging process in the power battery system.

[0020] Specifically, in this embodiment, optimizing the online current optimizer according to the multi-particle swarm algorithm includes: 1) Initialize the particle swarm parameters, set the total number of particles N to 100, the number of particles n in the main swarm and the four sub-swarms is 20, and set the maximum number of iterations T max is 200, inertia weight It decreases linearly from 0.9 to 0.4, and sets the learning factors C1 and C2 to 2; 2) Randomly generate particle positions and speed , and satisfy the parameter constraint adjustment, iterative optimization when the optimal fitness iteration change is less than 1e-6 or the maximum number of iterations t=T is reached max When , it is judged that the termination condition is reached; 3) Dynamic adjustment based on SOC and temperature , and according to , limit the current to output the optimal current value; 4) The output will be The current at the moment is the optimal current , the charger obtains the optimal current according to the optimizer Implement the charging process within the next fast charging cycle in the power battery system.

[0021] Embodiment 3 The capacity of the battery system is estimated and related parameters are updated based on the gated neural network.

[0022] Specifically, in this embodiment, the step of estimating the battery system capacity includes: 1) Obtain battery parameters such as voltage, current and temperature, set the sampling rate to 1Hz~10Hz, obtain time series such as timing data of charge and discharge cycles, calibrate actual capacity data through standard capacity test, and perform data cleaning and normalization; 2) Obtain the real-time values ​​of V, I, T and their first-order derivatives, including dV / dt, dI / dt, obtain statistical characteristics such as voltage fluctuation variance, current integral and temperature gradient, as well as cycle characteristics such as the current number of charge and discharge cycles and historical capacity attenuation rate: 3) Select a 5-30s time window based on the battery relaxation time, adjust the number of LSTM units according to the data complexity, the adjustment range is 64-256, set the Dropout rate to 0.2-0.5, and limit the output to the [0,1] interval according to the Sigmoid activation function. 4) Set the mean square error as the loss function and define the Adam optimizer, set the learning rate to 1e-4 and the decay rate to 1e-6, perform grid search by testing different time windows, and optimize the hyperparameters to build a battery capacity estimation model based on a gated neural network; 5) Divide the processed data set into training set, test set and validation set, train the model based on the training set, and evaluate the model through the test set and validation set. The evaluation indicators include: Absolute error: , Relative error: .

[0023] The beneficial effects of the present invention are: 1) by constructing a battery system model to consider the inconsistency of parameters between cells, it is possible to avoid reaching the boundary during the charging process of cells with individual parameter deviations, and to prevent lithium deposition, thermal runaway and other behaviors, thereby optimizing the problem of ignoring cell differences in traditional strategies, resulting in uneven fast charging current distribution and accelerated local lithium deposition / aging; 2) By estimating the health status and updating the model parameters (capacity and internal resistance) after battery aging, the problem of the aged battery system not being able to adapt to the original charging strategy can be avoided immediately, thereby optimizing the problem of the traditional strategy being unable to adapt to battery degradation, and the problem of the charging speed decreasing or the safety threshold being exceeded after battery aging; 3) The optimal fast charging control method for the battery system throughout its life cycle achieves online multi-objective lossless optimal fast charging within one charging cycle, while considering the threshold constraints of the internal negative electrode potential and external temperature to avoid other key side reaction problems such as lithium plating. It also combines a closed-loop observer and optimizer to dynamically optimize the current, which can optimize the problems that only consider the influence of unilateral factors in traditional methods.

[0024] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimal fast charging control of a battery system during its entire life cycle, characterized in that: It includes high-frequency fast charging optimization, medium-frequency lithium plating detection, and low-frequency parameter update. The specific steps are as follows: S1. Build an equivalent model based on the power battery system, accurately estimate the internal and external potential and temperature information of the battery, and achieve multi-objective optimal fast charging control within a fast charging cycle; S2. According to the optimal charging current, the power battery system performs lithium plating warning detection after the charge and discharge cycle; S3. The power battery system estimates the health status and updates the model parameters; S4. After the equivalent model is updated, optimal fast charging control of the battery system over its entire life cycle is achieved based on the obtained optimal charging current.

2. The method for optimal fast charging control of a battery system during its entire life cycle according to claim 1, characterized in that: The multi-objective optimal fast charging control of the power battery system in a fast charging cycle includes: 1) Establish a polarization equivalent circuit model based on the 3P6S module composed of monomer equivalent models, set model parameters and consider the parameter distribution between monomers; 2) Setting the objective function and constraints for optimal fast charging in the power battery system; 3) Construct an online state observer based on the equivalent model to automatically correct the voltage and current information estimated by the model at the next moment according to the BMS voltage and current signals collected during the charging process of the power battery system and the voltage and current information estimated by the model; 4) Build an online current optimizer based on the charging strategy, obtain the battery estimated state data according to the preset objective function and constraints and the real-time input state observer to obtain the optimal charging current at the next moment; 5) The charger obtains the optimal current at the next moment according to the online current optimizer to charge the power battery system.

3. The method for optimal fast charging control of a battery system during its entire life cycle according to claim 2, characterized in that: The specific steps of constructing the polarization equivalent circuit model include: 1) Set the polarization Rint model of thermoelectric coupling as the battery cell model, where the polarization Rint model includes the electrical model module and thermal model module of the battery cell. The model circuit principle is expressed as follows: ， ， in is the positive terminal voltage of the battery, is the negative terminal voltage of the battery, is the full battery terminal voltage, is the positive open circuit voltage of the battery, is the open circuit voltage of the negative electrode of the battery, is the full battery open circuit voltage, is the internal resistance of the positive electrode of the battery, is the internal resistance of the negative electrode of the battery, The current passing through the battery is set to the discharge current. Is a positive value, the current during charging is a negative value; 2) Based on the heat generation and heat dissipation principle of the lumped thermal model, a thermoelectric coupling model of the battery cell is constructed, which is expressed as: ， in is the heat generated by the battery per unit time, is the battery temperature, is the heat transfer per unit time, is the heat transfer coefficient, is the battery surface area, is the ambient temperature, then the battery temperature The expression of time variation is: ， in and Respectively represent the battery mass and battery specific heat capacity, and an equivalent model is established based on the thermoelectric coupling model; 3) Analyze the inconsistency of parameters between battery cells, including capacity , internal resistance and initial charge Inconsistency, set each parameter to conform to the normal distribution , then it is expressed as: ， in is the initial capacity of the battery, is the internal resistance change rate of battery model B, and its mean 1, initial charge Corresponds to the initial SOC distribution and varies with the initial state of the system. The larger it is, the greater the deviation of the parameter distribution.

4. The method for optimal fast charging control of a battery system during its entire life cycle according to claim 2, characterized in that: The objective function represents minimizing the time it takes for the power battery system to reach the target power level. The constraints represent the maximum and minimum limits on battery temperature, voltage, and charging current, including: 1) When the time required for the power battery system to reach the target state of charge is set to be the shortest within a charging cycle, the objective function is defined as: ， in Indicates the time Battery of the moment , The target power , set goals is 90%, Indicates the time The optimal current of the battery system at the moment; 2) Define the constraints as , The maximum charging current based on the 4C charging rate 232A, the maximum voltage of the battery cell 4.4V, the lowest voltage The lowest negative electrode potential threshold of the battery is 2.8V. 0.01V, the maximum battery temperature The lowest temperature is 50℃ -20℃, Estimate the negative electrode potential inside the battery at time k for the battery system.

5. The method for optimal fast charging control of a battery system during its entire life cycle according to claim 2, characterized in that: An online state observer based on an equivalent model is constructed based on an extended Kalman filter. The terminal voltage and temperature at time k estimated by the battery system model are compared with the terminal voltage and temperature at time k collected by the actual BMS, and the estimated terminal voltage, negative electrode potential and temperature at time k+1 are corrected through gain. The online current optimizer is optimized according to the multi-particle swarm algorithm, and the output is obtained as The current at the moment is the optimal current The charger uses the optimal current at the next moment obtained by the optimizer to realize the charging process in the power battery system.

6. The method for optimal fast charging control of a battery system during its entire life cycle according to claim 1, characterized in that: The lithium deposition early warning method is a relaxation voltage differential method, which specifically includes: 1) Set the data collection duration to 600s~1200s, and select any time period within this duration to collect the static voltage and time data of the battery after fast charging; 2) Perform time differential analysis on each collected single-cell voltage signal to determine whether there is a minimum value. The differential analysis expression is: ,in is the sampling frequency, is the voltage difference, Count the sampling points; when When there is no minimum value, it is judged that the battery system has no lithium deposition or has a slight lithium deposition process within the acceptable threshold, and the multi-objective optimal fast charging control operation is repeated for the next charging; when When it is a minimum value, it is judged that there is a lithium deposition side reaction in the battery system, and the health status of the battery system is estimated and the model parameters are updated, including: 1) Estimate the capacity of the battery system and update the parameters based on the gated neural network; 2) When the battery system is charged for the next moment, the internal resistance of the battery at the current temperature and power is calculated based on the initial charging current. The calculation formula is: , in is the initial current, is the battery voltage mutation difference caused by charging for 10 seconds, is the total internal resistance of the battery during 10s of charging, It is the total internal resistance of the battery under the same SOC and temperature conditions within 10 seconds of charging before aging. is the internal resistance change value; According to the change of internal resistance Update the battery internal resistance, specifically: ， in and are the internal resistance growth factors of the positive and negative electrodes, respectively, and , then set =0 and =1, thereby ensuring to the greatest extent that the updated fast charging strategy will not cause lithium plating at the negative electrode; 3) Based on the updated battery capacity and internal resistance parameter information, update the capacity, average value of positive electrode internal resistance and negative electrode internal resistance in the battery system model, and increase the parameter inconsistency between battery cells. The specific updates include: , according to the updated battery system model, charging is carried out in the next fast charging cycle based on the multi-objective optimal fast charging control method.

Images