Temperature control method for lithium ion power battery of new energy automobile

Abstract

A temperature control method for a lithium ion power battery of a new energy automobile comprises the steps that a lumped parameter thermal model of the lithium ion power battery is established, parameter identification is conducted on the lumped parameter thermal model through an eagle and African vulture mixed optimization algorithm, and model parameters are obtained; constructing a maximum correlation entropy tracking extended Kalman filter, and inputting the model parameters into the maximum correlation entropy tracking extended Kalman filter; a multi-target grey wolf algorithm improved based on a deep reinforcement learning algorithm is introduced to optimize the state noise covariance and the observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, so that the internal temperature of the battery is estimated; and establishing a model prediction controller based on the estimated internal temperature of the battery, improving the model prediction controller, and controlling the internal temperature of the battery through the improved model prediction controller. According to the invention, accurate online estimation and efficient control can be carried out on the internal temperature of the battery.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicles, and particularly to a temperature control method for lithium-ion power batteries of new energy vehicles. Background Art

[0002] With the rapid development of the new energy vehicle industry, the performance and safety of lithium-ion power batteries, as the core components of new energy vehicles, have attracted much attention. Temperature is a key factor affecting battery performance. Excessive temperature will affect the battery life and endurance, and even cause thermal runaway. With the continuous improvement of the energy density of lithium-ion batteries and the popularization of fast charging technology, the battery thermal management system urgently needs more efficient and accurate heat dissipation methods.

[0003] Liquid cooling technology has become the mainstream technology for thermal management due to its excellent heat dissipation performance and adaptability. However, traditional control strategies are difficult to balance heat dissipation efficiency and energy consumption optimization. Moreover, during the operation of the battery, the internal temperature is higher than the surface temperature and it is not easy to directly measure through sensing technology. Therefore, how to accurately estimate and efficiently control the internal temperature of the battery is a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the present invention provides a temperature control method for lithium-ion power batteries of new energy vehicles to accurately estimate and efficiently control the internal temperature of the battery.

[0005] A temperature control method for lithium-ion power batteries of new energy vehicles includes: Step S1: Establish a lumped parameter thermal model of the lithium-ion power battery, and use the hybrid optimization algorithm of Tianying and African vulture to identify the parameters of the lumped parameter thermal model to obtain the model parameters; Step S2: Based on the lumped parameter thermal model established in Step S1 and the obtained model parameters, construct a maximum correlation entropy strong tracking extended Kalman filter, and input the model parameters obtained in Step S1 into the maximum correlation entropy strong tracking extended Kalman filter; Step S3: Introduce a multi-objective grey wolf algorithm improved based on the deep reinforcement learning algorithm to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, so as to estimate the internal temperature of the battery; Step S4: Based on the internal temperature of the battery estimated in Step S3, establish a model predictive controller, use a black-winged kite optimization algorithm improved based on multiple strategies to optimize the prediction step and control step of the model predictive controller, and then introduce a temperature constraint relaxation adaptive strategy to adaptively adjust the weight matrix of the model predictive controller, so as to obtain an improved model predictive controller. Finally, control the internal temperature of the battery through the improved model predictive controller.

[0006] The temperature control method for lithium-ion power batteries of new energy vehicles provided by the present invention has the following beneficial effects: 1. Considering that the internal temperature of the lithium-ion power battery is higher than the surface temperature, the present invention first establishes a lumped parameter thermal model according to the battery thermal characteristics, and uses the hybrid optimization algorithm of Tianying and African vultures for parameter identification. The extended Kalman filter improved based on the maximum correlation entropy criterion and strong tracking filtering is used to estimate the internal temperature of the battery, and the multi-objective grey wolf algorithm improved based on deep reinforcement learning is used to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, improving the estimation accuracy of the filter. According to the estimated internal temperature of the battery, a model predictive controller is established. Through the temperature constraint relaxation adaptive strategy, the energy consumption of the system can be reduced on the premise of ensuring battery safety. The black-winged kite optimization algorithm improved based on multiple strategies is introduced to optimize the prediction step and control step of the model predictive controller, completing the improvement of the model predictive controller. Through the improved model predictive controller, the internal temperature of the battery can be controlled efficiently and accurately.

[0007] 2. Different from general battery thermal model parameter identification methods, the present invention uses the hybrid optimization algorithm of Tianying and African vultures for parameter identification, combining the high-altitude exploration ability of the Tianying algorithm and the group cooperation development ability of the African vulture algorithm to form a more efficient search mechanism. The hybrid algorithm of Tianying and African vultures can dynamically adjust the search strategy to adapt to the sensitivity differences of different parameters of the battery thermal model. Compared with other traditional algorithms, the hybrid algorithm of Tianying and African vultures is based on swarm intelligence agents and has stronger robustness to the noise of the objective function, making it more suitable for battery thermal management systems with uncertain noise.

[0008] 3. Different from general Kalman filters, the present invention constructs a maximum correlation entropy strong tracking extended Kalman filter by using the maximum correlation entropy criterion and fusing strong tracking filtering. Aiming at the problems of poor estimation accuracy and poor noise adaptability of the traditional extended Kalman filter in nonlinear systems, the present invention enhances the estimation accuracy of the filter by introducing the maximum correlation entropy criterion and strong tracking filtering. Finally, aiming at the problem that manual parameter tuning is relatively complex and difficult to achieve the optimal effect, the multi-objective grey wolf algorithm improved based on deep reinforcement learning is used to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, enabling accurate estimation of the internal temperature of the battery.

[0009] 4. Different from general model predictive controllers, the present invention introduces a temperature constraint relaxation adaptive strategy to adaptively adjust the weight matrix of the model predictive controller. To prevent the internal temperature of the battery from exceeding the threshold during operation, the temperature constraint relaxation adaptive strategy is adopted. When the battery temperature exceeds the safe range, the weight matrix is dynamically adjusted to give priority to controlling the internal temperature of the battery, sacrificing system energy consumption to ensure the safe operation of the battery. Finally, to improve the controller performance and enhance robustness, in addition, the present invention uses a black-winged kite optimization algorithm improved based on multiple strategies to optimize the prediction step and control step of the model predictive controller, introducing an adaptive attack strategy, a prey tracking strategy, and a strategy combining prey escape mechanism and trust region mutation strategy to improve the parameter optimization ability of the black-winged kite optimization algorithm. Description of the Drawings

[0010] Figure 1 It is a flowchart of the temperature control method for a lithium-ion power battery of a new energy vehicle provided by an embodiment of the present invention; Figure 2 It is a comparison chart of the estimated value of the internal temperature of a lithium-ion power battery by the method proposed by the present invention, the estimated value by the traditional extended Kalman filter algorithm, and the true value; Figure 3 It is a comparison chart of the internal temperature value of the battery after being controlled by the method proposed by the present invention, the internal temperature value of the battery after traditional fuzzy control, and the ideal internal temperature value of the battery. Detailed Embodiment

[0011] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions from beginning to end. The embodiments described below by referring to the drawings are exemplary and are intended to explain the embodiments of the present invention, and should not be construed as a limitation to the present invention.

[0012] Please refer to Figure 1 , an embodiment of the present invention provides a temperature control method for a lithium-ion power battery of a new energy vehicle, including steps S1 to S4: Step S1, establish a lumped parameter thermal model of the lithium-ion power battery, and use a hybrid optimization algorithm of a Tianying and an African vulture to identify the parameters of the lumped parameter thermal model to obtain the model parameters.

[0013] Among them, the expression of the established lumped parameter thermal model of the lithium-ion power battery is: ; ; ; Among them, and are the internal heat capacity and the external heat capacity of the battery respectively, and are the internal thermal resistance and the external thermal resistance respectively, is the value of the battery internal temperature minus the ambient temperature, , and are the battery internal temperature and the ambient temperature respectively, is the differential of, is the time the differential of, is the value of the battery surface temperature minus the ambient temperature, , is the battery surface temperature, is the differential of, is the heat generation amount of the battery during operation, is the average temperature of the battery, and are the open circuit voltage and the terminal voltage respectively, is the working current, is the entropy heat coefficient.

[0014] The process of parameter identification of the lumped parameter thermal model using the hybrid optimization algorithm of Tianying and African vulture is as follows: Step S11, determine the model parameters to be identified, including , , , ; Step S12, set the population size, the maximum number of iterations and the range of the parameters to be identified, generate the initial population, run the lumped parameter thermal model for each individual, and calculate the error of the individual using the first fitness function. The expression of the first fitness function is: ; Among them, is the true value of the battery surface temperature at time is the predicted value of the battery surface temperature output by the lumped parameter thermal model at time is the total time length; Step S13, select the leading vulture according to the fitness ranking, update the position of each vulture, and use the reflection method or random reset for the out-of-bounds parameters; Step S14, perform local refined search on the current optimal solution using the Tianying algorithm, update some individuals using the spiral path to enhance diversity; Step S15: When the maximum number of iterations is reached or the set error tolerance range is achieved, the algorithm terminates, completing parameter identification and obtaining the model parameters.

[0015] Step S2: Based on the lumped parameter thermal model established in Step S1 and the obtained model parameters, construct a maximum correlation entropy strong tracking extended Kalman filter, and input the model parameters obtained in Step S1 into the maximum correlation entropy strong tracking extended Kalman filter.

[0016] Among them, the process of constructing the maximum correlation entropy strong tracking extended Kalman filter is as follows: Step S21: Construct the system state equation, and the expression is: ; Among them, and are respectively at time and is the observation variable at time , and are respectively the state transition matrix, the control matrix, and the observation matrix, and are respectively the state noise at time and is the input variable at time Step S22: Introduce strong tracking filtering in the extended Kalman filter to obtain the strong tracking extended Kalman filter. The process is as follows: Define the innovation covariance matrix: ; Among them, and are respectively and the innovation covariance matrices at time is the innovation of strong tracking filtering at time denotes transpose, is the forgetting factor; Define the fading factor of strong tracking filtering, and the expression is: ; Among them, is the fading factor at time is the undetermined factor; Furthermore, obtain the covariance prediction, and the expression is: ; Among them, is the state noise covariance predicted according to at time and is the state noise covariance at time is the state noise covariance predicted at time is the estimated value of the state noise covariance at time Based on covariance prediction, a strong tracking extended Kalman filter is obtained; Step S23: During the operation of the battery, there is a large non-Gaussian noise interference. To avoid the influence of non-Gaussian noise, the maximum correlation entropy criterion is used to optimize the strong tracking extended Kalman filter, and a maximum correlation entropy strong tracking extended Kalman filter is obtained. The process is as follows: Convert the system state equation constructed in step S21 into the following formula: ; Among them, is the state variable predicted at time and is the joint covariance matrix of the state noise covariance and the observation noise covariance at time ; Perform Cholesky decomposition on : ; Among them, represents taking the expected value, is the observation noise covariance at time is the process variable corresponding to the covariance at time Multiply both sides of the transformed system state equation by to obtain the following formula: ; Among them, , , are all user-defined process variables, where: ; ; ; Define the cost function based on the maximum correlation entropy: ; Among them, is the cost function based on the maximum correlation entropy, is the kernel function, is the th element of is the th element of is the state dimension of; When is maximum, the optimal state estimate is obtained, and then the covariance modified by the maximum correlation entropy criterion is: ; where is the joint covariance matrix of the state noise covariance and the observation noise covariance at the moment modified by the maximum correlation entropy criterion, is the state noise covariance predicted according to the moment modified by the maximum correlation entropy criterion, at the is the observation noise covariance at the moment modified by the maximum correlation entropy criterion, is the process variable corresponding to the maximum correlation entropy criterion at the ; ; ; where and are intermediate matrices, represents a diagonal matrix; , , , respectively represent the 1st, th, th, th, th elements of , , , respectively represent the 1st, th, th, th, th elements of Furthermore, the state noise covariance and the modified observation noise covariance modified by the maximum correlation entropy criterion are obtained, and the expression is: ; ; where Modified according to the maximum correlation entropy criterion The observation noise covariance at the predicted time; Finally, the modified state noise covariance and the modified observation noise covariance are substituted into the strong tracking extended Kalman filter to obtain the maximum correlation entropy strong tracking extended Kalman filter.

[0017] Step S3: Introduce the multi-objective grey wolf algorithm improved based on the deep reinforcement learning algorithm to optimize the state noise covariance and the observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, so as to estimate the internal temperature of the battery.

[0018] Among them, introducing the multi-objective grey wolf algorithm improved based on the deep reinforcement learning algorithm to optimize the state noise covariance and the observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter specifically includes: Step S31: Set the size of the wolf pack, the maximum number of iterations, the structure of the deep reinforcement learning network, normalize the state, and standardize all observed values; Step S32: Calculate the fitness of each individual using the second fitness function, where the second fitness function is expressed as: ; Among them, and are respectively the true value and the estimated value of the state variable of the maximum correlation entropy strong tracking extended Kalman filter at the time; ; ; Among them, is the reward for adjusting the convergence factor of the grey wolf algorithm, is the number of iterations to reach the set convergence threshold, is the reward function of the weight coefficient of the grey wolf algorithm, and are respectively the fitness of the individual with the highest fitness in the th iteration and the th iteration; Step S34: When the maximum number of iterations or the convergence threshold is reached, the algorithm terminates, and the optimization of the state noise covariance and the observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter is completed.

[0019] Finally, the internal temperature of the battery is estimated by the maximum correlation entropy strong tracking extended Kalman filter.

[0020] Step S4: Based on the internal temperature of the battery estimated in step S3, a model predictive controller is established. The prediction step and control step of the model predictive controller are optimized by using the black-winged kite optimization algorithm improved based on multiple strategies. Then, a temperature constraint relaxation adaptive strategy is introduced to adaptively adjust the weight matrix of the model predictive controller, so as to obtain an improved model predictive controller. Finally, the internal temperature of the battery is controlled by the improved model predictive controller.

[0021] Among them, the process of optimizing the prediction step and control step of the model predictive controller by using the black-winged kite optimization algorithm improved based on multiple strategies specifically includes: The following multiple strategies are used to improve the black-winged kite optimization algorithm: The prey tracking strategy is introduced in the tracking stage of the black-winged kite optimization algorithm to improve the global search ability and convergence speed of the algorithm. The expression is: ; Among them, and are the -th and -th iteration of the tracking stage of the black-winged kite optimization algorithm, respectively, and the -dimensional position of the -th individual, is a random position, and are coefficients, , , is a non-linear factor, is a random number between 0 and 1; The adaptive attack strategy is introduced in the attack stage of the black-winged kite optimization algorithm to achieve the balance between global search and local search. The expression is: ； Among them, and are the -th and -th iteration of the attack stage of the black-winged kite optimization algorithm, respectively, and the -dimensional position of the -th individual, is a random number between 0 and 1, is a constant. In this embodiment, takes 0.85, and are variable parameters, , is the maximum number of iterations, ; In the migration stage of the Black-winged Kite Optimization Algorithm, a prey escape mechanism and a trust region mutation strategy are introduced to prevent the algorithm from falling into local optima and to increase the search intensity of the algorithm. The expression is as follows: ; where, and are the -th and -th iteration positions of the -th individual in the migration stage of the Black-winged Kite Optimization Algorithm, is the -dimensional position, and represent the prey escape energy and the prey escape energy threshold respectively, is the mean vector position of the trustworthy individuals, is the flight function, is a random number between 0 and 1; Then, based on the multi-strategy improved Black-winged Kite Optimization Algorithm, the prediction step and control step of the model predictive controller are optimized to obtain the objective function of the model predictive controller ; where, and are the prediction step and control step of the model predictive controller respectively, is the battery internal temperature value predicted by the model predictive controller at time based on the battery internal temperature estimated in step S3, is the set ideal temperature value, is the coolant flow rate change predicted by the model predictive controller at time for time, is the state error weight matrix of the model predictive controller at time, is the control input weight matrix. The safe temperature range of the battery is 15°C to 35°C. When the temperature is outside the safe range, a relaxed temperature constraint is selected to sacrifice system energy consumption to preferentially reduce the battery temperature. In the process of introducing a temperature constraint relaxation adaptive strategy to adaptively adjust the weight matrix of the model predictive controller, the following formula is satisfied:

[0022] ; ; where, is the state error weight matrix of the model predictive controller at time, and is the scaling factor, , , is the battery internal temperature predicted by the model predictive controller at time

[0023] Figure 2 Figure Figure 2 shows the comparison chart of the estimated value of the internal temperature of the lithium-ion power battery by the method proposed in the present invention, the estimated value of the traditional extended Kalman filter algorithm and the true value. It can be seen that the estimation accuracy of the estimation method proposed in the present invention for the internal temperature of the lithium-ion power battery is significantly improved, the internal temperature estimation error is within 0.231 °C, and the overall root mean square error is reduced by 60.7%. Figure 3 Figure Figure 3 shows the comparison chart of the internal temperature value of the battery controlled by the method proposed in the present invention, the internal temperature value of the battery after traditional fuzzy control and the ideal internal temperature value of the battery. It can be seen that compared with the fuzzy control method, the method proposed in the present invention significantly improves the convergence speed and control effect on the internal temperature of the lithium-ion power battery.

[0024] In summary, according to the temperature control method of the lithium-ion power battery of the new energy vehicle in the above embodiments, the following beneficial effects are obtained: 1. Considering that the internal temperature of the lithium-ion power battery is higher than the surface temperature, the present invention first establishes a lumped parameter thermal model according to the battery thermal characteristics, and uses the hybrid optimization algorithm of Tianying and African vulture to identify the parameters. The extended Kalman filter improved based on the maximum correlation entropy criterion and strong tracking filter is used to estimate the internal temperature of the battery, and the multi-objective grey wolf algorithm improved based on deep reinforcement learning is used to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, which improves the estimation accuracy of the filter. According to the estimated internal temperature of the battery, a model predictive controller is established. Through the temperature constraint relaxation adaptive strategy, the energy consumption of the system can be reduced on the premise of ensuring battery safety. The black-winged kite optimization algorithm improved based on multi-strategy is introduced to optimize the prediction step and control step of the model predictive controller, and the improvement of the model predictive controller is completed. Through the improved model predictive controller, the internal temperature of the battery can be controlled efficiently and accurately.

[0025] 2. Different from the general battery thermal model parameter identification method, the present invention uses the hybrid optimization algorithm of Tianying and African vulture to identify the parameters, combines the high-altitude exploration ability of the Tianying algorithm and the group cooperation development ability of the African vulture algorithm to form a more efficient search mechanism. The hybrid algorithm of Tianying and African vulture can dynamically adjust the search strategy to adapt to the sensitivity differences of different parameters of the battery thermal model. Compared with other traditional algorithms, the hybrid algorithm of Tianying and African vulture is based on swarm intelligence agents and has stronger robustness to the noise of the objective function, and is more suitable for the battery thermal management system with uncertain noise.

[0026] 3. Different from general Kalman filters, the present invention constructs a maximum correlation entropy strong tracking extended Kalman filter by adopting the maximum correlation entropy criterion and fusing strong tracking filtering. Aiming at the problems of poor estimation accuracy and poor noise adaptability of traditional extended Kalman filtering in nonlinear systems, the present invention enhances the estimation accuracy of the filter by introducing the maximum correlation entropy criterion and strong tracking filtering. Finally, aiming at the problem that manual parameter tuning is relatively complex and difficult to achieve the optimal effect, a multi-objective grey wolf algorithm improved based on deep reinforcement learning is used to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, which can achieve accurate estimation of the internal temperature of the battery.

[0027] 4. Different from general model predictive controllers, the present invention introduces a temperature constraint relaxation adaptive strategy to adaptively adjust the weight matrix of the model predictive controller. To prevent the internal temperature of the battery from exceeding the threshold during operation, the temperature constraint relaxation adaptive strategy is adopted. When the battery temperature exceeds the safe range, the weight matrix is dynamically adjusted to give priority to controlling the internal temperature of the battery, sacrificing system energy consumption to ensure the safe operation of the battery. Finally, to improve the performance of the controller and enhance robustness, in addition, the present invention uses a black-winged kite optimization algorithm improved based on multi-strategies to optimize the prediction step and control step of the model predictive controller, introduces an adaptive attack strategy, a prey tracking strategy, and combines a prey escape mechanism with a trust region mutation strategy to improve the black-winged kite optimization algorithm, thereby improving the parameter optimization ability.

[0028] The above embodiments only represent several implementation manners of the present invention, and the description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A temperature control method for a lithium-ion power battery of a new energy vehicle, characterized in that, Including: Step S1: Establish a lumped parameter thermal model of a lithium-ion power battery, and use a hybrid optimization algorithm of Tianying and African vulture to identify the parameters of the lumped parameter thermal model to obtain the model parameters; Step S2: Based on the lumped parameter thermal model established in Step S1 and the obtained model parameters, construct a maximum correlation entropy strong tracking extended Kalman filter, and input the model parameters obtained in Step S1 into the maximum correlation entropy strong tracking extended Kalman filter; Step S3: Introduce a multi-objective grey wolf algorithm improved based on the deep reinforcement learning algorithm to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, so as to estimate the internal temperature of the battery; Step S4: Based on the internal temperature of the battery estimated in Step S3, establish a model predictive controller, use a black-winged kite optimization algorithm improved based on multiple strategies to optimize the prediction step and control step of the model predictive controller, and then introduce a temperature constraint relaxation adaptive strategy to adaptively adjust the weight matrix of the model predictive controller, so as to obtain an improved model predictive controller. Finally, control the internal temperature of the battery through the improved model predictive controller.

2. The temperature control method of the lithium-ion power battery for new energy vehicles according to claim 1, characterized in that, In Step S1, the expression of the lumped parameter thermal model of the lithium-ion power battery established is: ； ； ； Wherein, and are the internal heat capacity and the external heat capacity of the battery respectively, and are the internal thermal resistance and the external thermal resistance respectively, is the value of the internal temperature of the battery minus the ambient temperature, , and are the internal temperature and the ambient temperature of the battery respectively, is 's differential, is time 's differential, is the value of the battery surface temperature minus the ambient temperature, , is the battery surface temperature, is 's differential, is the heat generation amount of the battery during operation, is the average temperature of the battery, and are the open circuit voltage and the terminal voltage respectively, is the working current, is the entropy heat coefficient.

3. The temperature control method for the lithium-ion power battery of a new energy vehicle according to claim 2, wherein In Step S1, the process of using a hybrid optimization algorithm of Tianying and African vulture to identify the parameters of the lumped parameter thermal model is: Step S11, determine the model parameters to be identified, including , , , ; Step S12, set the population size, the maximum number of iterations, and the range of parameters to be identified, generate an initial population, run the lumped parameter thermal model for each individual, and calculate the error of the individual using the first fitness function. The expression of the first fitness function is as follows: ； Wherein, is the true value of the battery surface temperature at the moment, is the predicted value of the battery surface temperature output by the lumped parameter thermal model at the moment, is the total time length; Step S13: Select the leading vulture according to the fitness ranking, update the position of each vulture, and use the reflection method or random reset for out-of-bounds parameters; Step S14: Use the Tianying algorithm to perform local refined search on the current optimal solution, and update some individuals using a spiral path to enhance diversity; Step S15: When the maximum number of iterations or the set error tolerance range is reached, the algorithm terminates, completes parameter identification, and obtains the model parameters.

4. The temperature control method of the lithium-ion power battery for new energy vehicles according to claim 3, characterized in that In Step S2, the process of constructing the maximum correlation entropy strong tracking extended Kalman filter is: Step S21: Construct a system state equation, the expression of which is: ； wherein, and are respectively the state variables at time and is the observation variable at time , and are respectively the state transition matrix, the control matrix and the observation matrix, and are respectively the state noise at time and is the input variable at time; Step S22: Introduce strong tracking filtering in the extended Kalman filter to obtain a strong tracking extended Kalman filter. The process is as follows: Define the innovation covariance matrix: ； Among them, and are respectively and the innovation covariance matrices at times, is the innovation of strong tracking filtering at time, denotes transpose, is the forgetting factor; Define the fading factor of strong tracking filtering, the expression of which is: ； Among them, is the fading factor at the moment, is a factor to be determined; Furthermore, obtain the covariance prediction, the expression of which is: ； Among them, is the state noise covariance predicted at time, is the state noise covariance predicted at time, and is the estimated value of the state noise covariance at Based on the covariance prediction, obtain the strong tracking extended Kalman filter; Step S23: Use the maximum correlation entropy criterion to optimize the strong tracking extended Kalman filter to obtain the maximum correlation entropy strong tracking extended Kalman filter. The process is as follows: Convert the system state equation constructed in Step S21 into the following form: ； Among them, is the state variable predicted at time is the joint covariance matrix of the state noise covariance and the observation noise covariance at For perform Cholesky decomposition: ； Among them, denotes taking the expected value, is the observation noise covariance at the moment, is the process variable corresponding to the covariance at the moment; Multiply both sides of the transformed system state equation by to obtain the following equation: ； Among them, , , are all user-defined process variables, where: ； ； ； Define the cost function based on the maximum correlation entropy: ； Among them, is the cost function based on the maximum correlation entropy, is the kernel function, is the -th element in is the -th element in is the state dimension of When is maximized, the optimal state estimate is obtained, and then the covariance modified by the maximum correlation entropy criterion is: ； Among them, is the joint covariance matrix of the state noise covariance and the observation noise covariance after being modified by the maximum correlation entropy criterion, and is the state noise covariance predicted at time according to the maximum correlation entropy criterion after modification, and is the observation noise covariance at time after being modified by the maximum correlation entropy criterion, and is the process variable corresponding to the maximum correlation entropy criterion at ； ； ； Among them, and are intermediate matrices, represents a diagonal matrix; , , , respectively represent the 1st, the th, the th, the th elements; , , , respectively represent the 1st, the th, the th, the th elements; Furthermore, obtain the modified state noise covariance and modified observation noise covariance after the maximum correlation entropy criterion, the expression of which is: ； ； Among them, is the observation noise covariance at time predicted according to the maximum correlation entropy criterion modified at time; Finally, substitute the modified state noise covariance and modified observation noise covariance into the strong tracking extended Kalman filter to obtain the maximum correlation entropy strong tracking extended Kalman filter.

5. The temperature control method of the lithium-ion power battery for new energy vehicles according to claim 4, wherein, In step S3, a multi-objective grey wolf algorithm improved based on a deep reinforcement learning algorithm is introduced to optimize the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter, which specifically includes: Step S31, set the wolf pack size, maximum number of iterations, deep reinforcement learning network structure, normalize the state, and standardize all observation values; Step S32, calculate the fitness of each individual using the second fitness function, where the second fitness function has the following expression: ； wherein, and are respectively the true value and the estimated value of the state variable of the maximum correlation entropy strong tracking extended Kalman filter at a moment; Step S33, establish a reward function for the deep reinforcement learning algorithm, and use the reward function to update the convergence factor and weight coefficient of the grey wolf algorithm. The established reward function is: ； ； Among them, is the reward for adjusting the convergence factor of the Grey Wolf algorithm, is the number of iterations to reach the set convergence threshold, is the reward function for the weight coefficient of the Grey Wolf algorithm, and are respectively the fitness of the individual with the highest fitness in the -th iteration and the -th iteration; Step S34, when the maximum number of iterations or the convergence threshold is reached, the algorithm terminates, and the optimization of the state noise covariance and observation noise covariance of the maximum correlation entropy strong tracking extended Kalman filter is completed.

6. The temperature control method for the lithium-ion power battery of a new energy vehicle according to claim 1, wherein In step S4, the process of optimizing the prediction step and control step of the model predictive controller by using a black-winged kite optimization algorithm improved based on multiple strategies specifically includes: The following multiple strategies are used to improve the black-winged kite optimization algorithm: Introduce a prey tracking strategy in the tracking stage of the black-winged kite optimization algorithm, and the expression is: ； Among them, and are respectively the -th iteration and the -th iteration of the tracking stage of the Black-winged Kite Optimization Algorithm, and are the -dimensional positions of the -th individual, is a random position, and are coefficients, , , are non-linear factors, is a random number between 0 and 1; Introduce an adaptive attack strategy in the attack stage of the black-winged kite optimization algorithm to achieve the balance between global search and local search, and the expression is: ； Among them, and are the -th iteration and the -th iteration in the attack phase of the Black-winged Kite Optimization Algorithm for the -th individual's -dimensional position, is a random number between 0 and 1, is a constant, and are variable parameters, , is the maximum number of iterations, ; In the migration stage of the black-winged kite optimization algorithm, enter the prey escape mechanism and trust region mutation strategy, and the expression is: ； Among them, and are respectively the -th iteration and the -th iteration of the migration stage of the Black-winged Kite Optimization Algorithm, and the -dimensional position of the -th individual, and represent the prey escape energy and the prey escape energy threshold respectively, is the mean vector position of the trustworthy individuals, is the flight function, is a random number between 0 and 1; Then, the prediction step and control step of the model predictive controller are optimized based on the multi-strategy improved Elanus caeruleus optimization algorithm to obtain the objective function of the model predictive controller as follows: ； Among them, and are the prediction step and the control step of the model predictive controller respectively, is the battery internal temperature value predicted by the model predictive controller based on the battery internal temperature at the moment estimated in step S3 at the moment at the moment, is the set ideal temperature value, is the coolant flow rate change predicted by the model predictive controller at the moment at the moment, is the state error weight matrix of the model predictive controller at the moment, is the control input weight matrix.

7. The temperature control method of the lithium-ion power battery for new energy vehicles according to claim 6, characterized in that, In step S4, when introducing a temperature constraint relaxation adaptive strategy to adaptively adjust the weight matrix of the model predictive controller, the following formula is satisfied: ； Among them, is the state error weight matrix of the model predictive controller at time, and are scale factors, , , is the predicted internal battery temperature by the model predictive controller at time.

Images