Thermal control method and device of solid-state battery, electronic equipment and readable storage medium

Abstract

The invention provides a thermal control method and device for a solid-state battery, electronic equipment and a readable storage medium, and the method comprises the steps: obtaining the real-time state data of a target solid-state battery; inputting the real-time state data into the trained reinforcement learning model to obtain the parameter variation of the proportional integral differential controller and the target temperature offset of the target solid-state battery; and based on the parameter variation and the target temperature offset, adjusting an output duty ratio of a proportional-integral-derivative controller so as to adjust the heating power of a heating film of the target solid-state battery. Therefore, the control parameters of the proportional integral differential controller are dynamically optimized in real time through the trained reinforcement learning model, the control precision of the solid-state battery thermal control is ensured, the response time of the solid-state battery thermal control is shortened, and the control efficiency of the solid-state battery thermal control is improved.

Description

Technical Field

[0001] This application relates to the field of solid-state battery thermal control technology, and in particular to thermal control methods, devices, electronic devices, and readable storage media for solid-state batteries. Background Technology

[0002] In traditional battery thermal management systems, proportional-integral-derivative (PID) control is widely used due to its simple structure, high reliability, and lack of reliance on accurate models.

[0003] However, the inherent shortcomings of traditional PID controllers are amplified in the application scenarios of solid-state batteries, which are extremely sensitive to temperature. Specifically, the parameters of traditional PID controllers are usually fixed values, making it difficult to cope with sudden changes in ambient temperature and sudden changes in battery heat generation power. Furthermore, the thermal dynamic response of batteries is characterized by high inertia and nonlinearity, and the ionic conductivity of solid electrolytes is very sensitive to temperature. Traditional PID control can lead to temperature regulation lag, overshoot, and even oscillation. Large temperature fluctuations directly affect the performance and lifespan of solid-state batteries, resulting in low control accuracy and long response time in solid-state battery thermal control, and consequently, low control efficiency. Summary of the Invention

[0004] In view of this, embodiments of this application provide at least a thermal control method, apparatus, electronic device, and readable storage medium for solid-state batteries. By using a trained reinforcement learning model to dynamically optimize the control parameters of the proportional-integral-derivative controller in real time, the control accuracy of solid-state battery thermal control is ensured while reducing the response time of solid-state battery thermal control and improving the control efficiency of solid-state battery thermal control.

[0005] This application mainly includes the following aspects: In a first aspect, embodiments of this application provide a thermal control method for a solid-state battery, the method comprising: Acquire real-time status data of the target solid-state battery; The real-time state data is input into the trained reinforcement learning model to obtain the parameter changes of the proportional-integral-derivative controller and the target temperature offset of the target solid-state battery. Based on the parameter changes and the target temperature offset, the output duty cycle of the proportional-integral-derivative controller is adjusted to adjust the heating power of the heating film of the target solid-state battery.

[0006] Secondly, embodiments of this application also provide a thermal control device for a solid-state battery, the thermal control device for the solid-state battery comprising: The data acquisition module is used to acquire real-time status data of the target solid-state battery; The reinforcement learning module is used to input the real-time state data into the trained reinforcement learning model to obtain the parameter changes of the proportional-integral-derivative controller and the target temperature offset of the target solid-state battery. The heating control module is used to adjust the output duty cycle of the proportional-integral-derivative controller based on the parameter change and the target temperature offset, so as to adjust the heating power of the heating film of the target solid-state battery.

[0007] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory through the bus. The machine-readable instructions are executed by the processor to perform the steps of the thermal control method for solid-state batteries as described above.

[0008] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the thermal control method for a solid-state battery as described above.

[0009] The solid-state battery thermal control method, apparatus, electronic device, and readable storage medium provided in this application include: acquiring real-time state data of the target solid-state battery; inputting the real-time state data into a trained reinforcement learning model to obtain the parameter change of the proportional-integral-derivative (PID) controller and the target temperature offset of the target solid-state battery; and adjusting the output duty cycle of the PID controller based on the parameter change and the target temperature offset to adjust the heating power of the heating film of the target solid-state battery. In this way, by dynamically optimizing the control parameters of the PID controller in real time through the trained reinforcement learning model, the control accuracy of the solid-state battery thermal control is ensured while reducing the response time and improving the control efficiency.

[0010] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1A flowchart of a thermal control method for a solid-state battery provided in an embodiment of this application is shown; Figure 2 This invention provides a functional block diagram of a thermal control device for a solid-state battery according to an embodiment of the present application. Figure 3 This is a second functional block diagram of a thermal control device for a solid-state battery provided in an embodiment of this application; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0014] It is worth noting that prior to this application, existing solutions typically used fixed-value PID controllers, which struggled to handle sudden changes in ambient temperature and abrupt shifts in battery heat generation. Furthermore, the thermal dynamic response of batteries exhibits high inertia and nonlinearity, and the ionic conductivity of solid-state electrolytes is highly sensitive to temperature. Traditional PID control can lead to temperature regulation lag, overshoot, and even oscillations. Large temperature fluctuations directly impact the performance and lifespan of solid-state batteries. Moreover, battery thermal management is a coupled multiple-input multiple-output (MIMO) system where coolant temperature, ambient temperature, and state of charge (SOC) interact, making it difficult for traditional PID control to efficiently coordinate these variables.

[0015] Other solutions employ end-to-end model control, directly outputting PWM signals to control the heating film to heat the battery by collecting information such as battery temperature. However, results show that the model learning efficiency is low, highly dependent on the completeness of the training data scenario, has a heavy online computational burden, and is extremely complex in design and debugging, making it difficult for end-to-end control to achieve good results in highly coupled battery thermal management systems. Solid-state battery thermal control has low control precision and long response time, which in turn leads to low control efficiency.

[0016] To address the aforementioned issues, embodiments of this application provide a thermal control method, apparatus, electronic device, and readable storage medium for solid-state batteries. By using a trained reinforcement learning model to dynamically optimize the control parameters of the proportional-integral-derivative controller in real time, the control accuracy of solid-state battery thermal control is ensured while reducing the response time and improving the control efficiency of solid-state battery thermal control.

[0017] To facilitate understanding of this application, the technical solutions provided in this application will be described in detail below with reference to specific embodiments.

[0018] Please see Figure 1 , Figure 1 This is a flowchart illustrating a thermal control method for a solid-state battery provided in an embodiment of this application. Figure 1 As shown in the embodiment of this application, the thermal control method for solid-state batteries includes the following steps: S101, acquire real-time status data of the target solid-state battery.

[0019] In this embodiment, a sensor network deployed in the solid-state battery system collects real-time status data of the target solid-state battery, including parameters such as battery temperature, ambient temperature, and coolant temperature. Simultaneously, it obtains operational data such as battery state of charge and charge / discharge rate from the battery management system. This real-time status data is transmitted to the central processing unit via the system bus, providing a data foundation for subsequent intelligent control decisions.

[0020] S102, the real-time state data is input into the trained reinforcement learning model to obtain the parameter change of the proportional-integral-derivative controller and the target temperature offset of the target solid-state battery.

[0021] In this embodiment, preprocessed real-time state data is input into a reinforcement learning model pre-trained based on an advantage-weighted actor-critic algorithm. The model's actor network outputs parameter adjustment instructions, including changes in proportional coefficients, integral coefficients, and derivative coefficients, as well as a target temperature offset, based on the input state data. These outputs serve as optimization instructions, dynamically adjusting the operating parameters of the underlying proportional-integral-derivative (PID) controller.

[0022] S103, based on the parameter change and the target temperature offset, adjust the output duty cycle of the proportional-integral-derivative controller to adjust the heating power of the heating film of the target solid-state battery.

[0023] In this embodiment, the control parameters of the proportional-integral-derivative (PID) controller are updated in real time based on the parameter changes output by the reinforcement learning model, while the temperature setpoint is adjusted according to the target temperature offset. The updated PID controller, based on the new parameter settings and temperature setpoint, and combined with the real-time acquired battery temperature, calculates the optimal pulse-width modulation (PWM) duty cycle signal and outputs this signal to the heating film drive circuit to precisely control the heating power of the heating film, thereby achieving precise temperature control of the solid-state battery.

[0024] Furthermore, before acquiring real-time state data of the target solid-state battery, the reinforcement learning model is trained according to the following steps: Step a1: Establish a solid-state battery thermal simulation model, and conduct multi-condition simulation experiments based on the solid-state battery thermal simulation model to obtain multi-condition simulation generation data.

[0025] Here, a high-fidelity thermal simulation model of a solid-state battery is established using professional system simulation software. This model can accurately simulate the thermal dynamic characteristics of solid-state batteries under different operating conditions, including the battery's heat generation and dissipation processes, as well as its heat exchange behavior with the cooling system. By setting different combinations of parameters such as ambient temperature (covering -30℃ to 50℃), battery state of charge (SOC) (0%-100%), discharge rate (1C-2C), and target temperature (0℃ to 45℃), multiple batches of simulation experiments are conducted to generate simulation data containing multi-dimensional parameters such as battery temperature, coolant temperature, and ambient temperature, providing a rich data foundation for reinforcement learning training.

[0026] In this embodiment, AMESim software is used to establish a high-fidelity solid-state battery thermal simulation model to accurately simulate the heat generation characteristics of solid-state batteries, the response of the thermal management system, and the influence of the external environment.

[0027] Step a2: Based on the multi-condition simulation data and the historical operating data of the solid-state battery, determine the state space input of the reinforcement learning model.

[0028] Here, training data is generated through multi-condition simulation experiments, and real-world scenario data is collected. The simulation-generated data is then integrated and aligned with historical data accumulated during the actual operation of the solid-state battery system to construct a complete training dataset. Based on this dataset, the state space input of the reinforcement learning model is determined and formatted into a Dataset format, serving as the training data for the reinforcement learning model.

[0029] Step a3: Based on the state space input, the reinforcement learning model is trained using the advantage-weighted actor-critic algorithm to obtain the trained reinforcement learning model.

[0030] Here, the advanced offline reinforcement learning algorithm Advantage-Weighted Actor-Critic (AWAC) is employed for model training. The actor network is responsible for generating action policies based on the input state space, outputting parameter adjustment instructions including changes in proportional coefficients, integral coefficients, and derivative coefficients, as well as the target temperature offset. The critic network is responsible for evaluating the value of state-action pairs and guiding the actor network's updates through the advantage-weighted mechanism. A multi-objective reward function is used during training, comprehensively considering multiple optimization objectives such as temperature control accuracy, energy efficiency, and safety constraints. Through iterative training on a large amount of offline data, the model learns the optimal parameter adjustment strategies under different system states, ultimately obtaining a convergent, online-deployable reinforcement learning model, which is more suitable for training battery thermal management reinforcement models on offline static data.

[0031] Furthermore, the state space input of the reinforcement learning model includes battery state, environmental state, and target state and constraints; the battery state includes state of charge, battery temperature, battery temperature error, and charge / discharge rate; the environmental state includes coolant temperature, ambient temperature, and coolant flow rate; the target state and constraints include desired temperature, maximum allowable temperature rise rate, and safe temperature threshold.

[0032] In this embodiment, the three dimensions of the state space input together form the complete foundation for the reinforcement learning model's perception and control decisions. The battery state dimension provides the battery's own operating status, including the state of charge. This reflects the battery's remaining energy level and directly affects the battery's internal resistance and heat generation characteristics; battery temperature As the core controlled variable in thermal control, it is related to the battery temperature error. The difference between the set temperature and the actual temperature constitutes the key input for proportional-integral-derivative (PID) control; simultaneously, the charge / discharge rate... As a crucial parameter, it determines the battery's instantaneous heat generation power. The environmental state dimension describes the system's external operating conditions, including coolant temperature. and ambient temperature These parameters constitute the system's thermal boundary conditions, as well as the coolant flow rate. This active thermal management system has controllable variables. The target state and constraint dimensions define the desired control objective and safety boundaries, with the desired temperature being... The target value for thermal control and the maximum permissible rate of temperature rise are provided. and safe temperature threshold Together, they ensured that the battery operated within safe limits. This multi-dimensional state-space design enabled the reinforcement learning model to fully perceive the system's operating state, comprehensively consider performance goals and safety constraints, and thus make more intelligent and safer control decisions.

[0033] In this embodiment of the application, the state space input It can be represented as: .

[0034] Further, the step of training the reinforcement learning model based on the state space input using the advantage-weighted actor-critic algorithm to obtain the trained reinforcement learning model includes: Step b1: The state space input is used as the input to the actor network, and the action vector is output according to the preset action rules of the actor network; the action vector includes the change of the proportional coefficient, the change of the integral coefficient, the change of the differential coefficient, and the target temperature offset.

[0035] In this embodiment, the Actor-Critic architecture employs the Adam neural network optimizer, which is highly sensitive to the scale of the input data. If the numerical ranges of the various dimensions of the state space input differ significantly, gradients may not converge quickly. Normalization stabilizes the data within a reasonable range, providing the network with a relatively stable input distribution. This facilitates better learning and generalization, leading to faster model convergence, improved training efficiency, and resolution of gradient explosion or vanishing problems. Therefore, the state space input is first normalized to obtain the normalized state space input.

[0036] Specifically, the normalized state of charge It can be represented as: .

[0037] Normalized battery temperature It can be represented as: .

[0038] Normalized battery temperature error It can be represented as: .

[0039] Normalized coolant temperature It can be represented as: .

[0040] Normalized ambient temperature It can be represented as: .

[0041] Normalized expected temperature It can be represented as: .

[0042] Maximum allowable temperature rise rate after normalization It can be represented as: .

[0043] Normalized safe temperature threshold It can be represented as: .

[0044] Normalized charge / discharge rate It can be represented as: .

[0045] Normalized coolant flow rate It can be represented as: .

[0046] Finally, the normalized state space input It can be represented as: .

[0047] In this embodiment of the application, the actor network receives the normalized state space input. It extracts deep features of the current system state through its internal multi-layer neural network structure. Based on the learned state features, the actor network generates corresponding action vectors according to preset action rules. This vector specifically includes the adjustment amounts for the three parameters of the proportional-integral-derivative (PID) controller (proportional, integral, and derivative coefficients) and the offset of the target temperature. These output quantities serve as optimization instructions for the control system, providing precise guidance for subsequent parameter adjustments. The parameter change method, which requires changes in the decision-making parameters of the reinforcement learning model, results in relatively smooth parameter changes, avoiding large oscillations in the simulation. Specifically, the action vector... It can be represented as: .

[0048] Step b2: The state space input and the action vector are used together as input to the critic network, and a value assessment is output according to the preset reward function of the critic network.

[0049] Here, the critic network simultaneously receives the current state space input and the action vectors generated by the actor network. The critic network evaluates the long-term value of the state-action pair using its independent neural network structure. The critic network calculates the value based on a pre-defined multi-objective reward function, which comprehensively considers multiple performance indicators such as temperature control accuracy, energy efficiency, and safety constraints, outputting a comprehensive value assessment score. This score reflects the long-term benefit of taking the action in the current state, providing guidance for the optimization of the actor network.

[0050] Step b3: Adjust the actor network according to the value assessment, iteratively optimize the reinforcement learning model until the model converges, and obtain the trained reinforcement learning model.

[0051] Here, based on the value assessment provided by the critic network, an advantage-weighted mechanism is used to update the parameters of the actor network. This mechanism, by weighting the advantage function, makes the model more focused on learning behavioral strategies that can bring significant performance improvements. During iterative training, the actor network and the critic network cooperate and co-evolve, gradually improving the performance of the control strategy through repeated learning on a large amount of training data, until the model strategy tends to stabilize and all performance indicators meet the preset requirements, ultimately obtaining a well-trained reinforcement learning model.

[0052] Furthermore, the preset action rules include: Based on the range of the absolute value of the battery temperature error, set the corresponding changes in the proportional coefficient, integral coefficient, and derivative coefficient.

[0053] Set the corresponding target temperature offset based on the range of the state of charge.

[0054] In this embodiment, the preset action rules adopt expert rules based on temperature error and the rate of change of temperature error, which are used to guide the training process of the reinforcement learning model.

[0055] Specifically, when the battery temperature error When the absolute value is large (e.g., exceeding 5°C), to avoid significant steady-state errors in the system and to enable the system to quickly approach the desired state, such as when the battery temperature is too low and needs to be rapidly increased, for example: , , .

[0056] Battery temperature error When the temperature is within a moderate range (e.g., 2°C to 5°C), the system exhibits a small error, which can be adjusted... To eliminate residual errors, the battery temperature is expected to reach the ideal temperature. For example: , , .

[0057] Battery temperature error At lower temperatures (e.g., below 2°C), the system requires fine-tuning, prioritizing accuracy while suppressing overshoot. For example: , , .

[0058] Battery temperature error Smaller (e.g., below 1°C) and At that time, it should be reduced significantly. At the same time increase This prevents system overshoot. For example: , , .

[0059] When the state of charge At that time, in order to compensate for the effect of increased battery internal resistance at low temperatures, .

[0060] In addition, the rules also set the action constraints for the action vectors. Specifically, , , , .

[0061] Furthermore, the preset reward function includes penalty terms corresponding to temperature error constraints, energy consumption constraints, temperature constraints, temperature rise rate constraints, charge / discharge rate constraints, and heat dissipation efficiency constraints.

[0062] In this embodiment, the preset reward function guides the reinforcement learning model to learn a control strategy that meets multi-objective optimization requirements through carefully designed penalty terms. Specifically, the temperature error constraint penalty term is constructed based on the absolute value of the deviation between the actual temperature and the target temperature, aiming to guide the model to prioritize the accuracy of temperature control; the energy consumption constraint penalty term is based on the power consumption of the heating film, prompting the model to minimize energy consumption while meeting temperature requirements; the temperature constraint penalty term imposes severe penalties when the battery temperature exceeds the safety threshold, ensuring that the battery always operates within a safe temperature range; the temperature rise rate penalty term constrains the rate of temperature rise exceeding the maximum allowable value, preventing rapid temperature changes from damaging battery life; the charge / discharge rate penalty term optimizes the battery's operating state by penalizing high-rate operation; and the heat dissipation efficiency penalty term is based on the operating state of the cooling system, promoting the improvement of system heat dissipation efficiency. These penalty terms are weighted and combined to form a comprehensive reward signal, enabling the reinforcement learning model to autonomously learn the optimal control strategy during training that balances energy consumption efficiency, safety constraints, and system stability while ensuring control accuracy.

[0063] Specifically, the reward function It can be represented as: .

[0064] in, This represents the operating power of the heating film and its energy consumption. , where is the error weight; Energy consumption weight; This is for over-temperature weighting; , which is the weight for the heating rate; , which represents the charge / discharge rate weighting; This represents the weight for heat dissipation efficiency. The constraints also include temperature safety constraints. and temperature rise rate constraints .

[0065] Further, adjusting the output duty cycle of the proportional-integral-derivative controller based on the parameter change and the target temperature offset to adjust the heating power of the heating film of the target solid-state battery includes: Step c1: Update the proportional coefficient, integral coefficient, and derivative coefficient of the proportional-integral-derivative controller according to the parameter changes; the parameter changes include the changes in the proportional coefficient, integral coefficient, and derivative coefficient.

[0066] Here, the system employs an incremental update strategy, superimposing the parameter changes output by the reinforcement learning model with the current parameter values ​​of the proportional-integral-derivative (PID) controller to calculate new control parameters in real time. Specifically, the new proportional coefficient equals the sum of the original proportional coefficient and its change; the new integral coefficient equals the sum of the original integral coefficient and its change; and the new derivative coefficient equals the sum of the original derivative coefficient and its change. This dynamic parameter adjustment mechanism enables the PID controller to adaptively optimize its control characteristics, employing the most suitable control parameters for different system states.

[0067] Step c2: Update the temperature setpoint according to the target temperature offset; wherein the updated temperature setpoint is the sum of the initial target temperature and the target temperature offset.

[0068] Here, the system obtains the optimized temperature setpoint by adding the initial target temperature to the target temperature offset output by the reinforcement learning model. This mechanism enables the system to intelligently adjust the temperature control target according to actual operating conditions. For example, when the battery's state of charge is low, the target temperature can be appropriately increased to compensate for the performance loss caused by the increase in battery internal resistance, thereby improving the overall performance of the system while ensuring safety.

[0069] Specifically, temperature set point .

[0070] Step c3: Based on the updated proportional coefficient, integral coefficient, derivative coefficient, and temperature setpoint, as well as the real-time battery temperature, the duty cycle of the proportional-integral-derivative controller is calculated.

[0071] Here, the updated proportional-integral-derivative controller is based on a new temperature setpoint and real-time battery temperature data. Calculate the current battery temperature error Then, combining the updated control parameters, the optimal pulse width modulation duty cycle is calculated using a position-based proportional-integral-derivative algorithm. This calculation process fully considers the real-time status of the system and the optimized control parameters to ensure that the output duty cycle can achieve fast, accurate, and stable temperature control.

[0072] Specifically, battery temperature error .

[0073] Duty cycle .

[0074] Step c4: The duty cycle of the proportional-integral-derivative controller is output to the heating film drive circuit of the target solid-state battery to adjust the heating power of the heating film.

[0075] Here, the calculated pulse width modulation duty cycle signal is transmitted to the heating film driving circuit through a digital output interface. The driving circuit adjusts the average voltage applied to the heating film based on the received duty cycle signal, thereby precisely controlling the heating power of the heating film. This pulse width modulation-based power control method has the advantages of fast response speed and high control accuracy, enabling precise management of the solid-state battery temperature and forming a complete closed-loop control system.

[0076] Specifically, the operating power of the heating film It can be represented as: .

[0077] in, This refers to the power supply voltage. This refers to the heating film resistor.

[0078] In this embodiment, under specific conditions—a battery initial state of 97% SOC, a battery temperature and an ambient temperature of -30°C, a charge / discharge rate of 1C, and a target heating temperature of 0°C—the traditional control method and the reinforcement learning control method exhibit significant differences in heating performance. The traditional control scheme takes 7 minutes to complete the heating process, with a total system energy consumption of 5 kWh and a temperature control accuracy fluctuation range of ±2°C. In contrast, the reinforcement learning scheme completes heating in just 5 minutes, reducing total energy consumption to 3.5 kWh, while significantly narrowing the temperature fluctuation range to ±0.5°C, demonstrating its comprehensive advantages in heating efficiency, energy economy, and control stability.

[0079] Thus, this application's embodiments address the shortcomings of traditional control algorithms in terms of poor adaptability, while also overcoming the problems of poor end-to-end control performance and difficulties in formal verification associated with deep learning algorithms. A strategy based on reinforcement learning to optimize PID parameters and target temperature is designed, leveraging reinforcement learning's strength in handling nonlinear and strongly coupled systems. This intelligently adjusts the PID controller, enabling it to dynamically adapt to complex and changing environments, consistently maintaining optimal control performance while balancing performance, safety, and energy efficiency, achieving smoother and more precise temperature control and higher overall performance. An offline training method is employed, using extensive real-world scenario data and AMESim model simulation results to train the reinforcement learning model offline, avoiding the excessively long training time caused by online co-training with the AMESim model.

[0080] This application provides a thermal control method for solid-state batteries, comprising: acquiring real-time state data of a target solid-state battery; inputting the real-time state data into a trained reinforcement learning model to obtain the parameter changes of a proportional-integral-derivative (PID) controller and the target temperature offset of the target solid-state battery; and adjusting the output duty cycle of the PID controller based on the parameter changes and the target temperature offset to adjust the heating power of the heating film of the target solid-state battery. In this way, by dynamically optimizing the control parameters of the PID controller in real time through the trained reinforcement learning model, the control accuracy of the solid-state battery thermal control is maintained while reducing the response time and improving the control efficiency.

[0081] Based on the same application concept, this application also provides a solid-state battery thermal control device corresponding to the solid-state battery thermal control method provided in the above embodiments. Since the principle of the device in this application is similar to the solid-state battery thermal control method in the above embodiments of this application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0082] Please see Figure 2 , Figure 2This is one of the functional block diagrams of a thermal control device for a solid-state battery provided in an embodiment of this application. For example... Figure 2 As shown, the thermal control device 200 for a solid-state battery provided in this application embodiment includes: The data acquisition module 210 is used to acquire real-time status data of the target solid-state battery.

[0083] The reinforcement learning module 220 is used to input the real-time state data into the trained reinforcement learning model to obtain the parameter changes of the proportional-integral-derivative controller and the target temperature offset of the target solid-state battery.

[0084] The heating control module 230 is used to adjust the output duty cycle of the proportional-integral-derivative controller based on the parameter change and the target temperature offset, so as to adjust the heating power of the heating film of the target solid-state battery.

[0085] Further, please refer to Figure 3 , Figure 3 This is a second functional block diagram of a thermal control device for a solid-state battery provided in an embodiment of this application. Figure 3 As shown, the thermal control device 200 for solid-state batteries also includes: The data generation module 240 is used to establish a solid-state battery thermal simulation model and conduct multi-condition simulation experiments based on the solid-state battery thermal simulation model to obtain multi-condition simulation generation data.

[0086] The input determination module 250 is used to determine the state space input of the reinforcement learning model based on the multi-condition simulation generated data and the historical operating data of the solid-state battery.

[0087] The model training module 260 is used to train the reinforcement learning model based on the state space input using the advantage-weighted actor-critic algorithm, so as to obtain the trained reinforcement learning model.

[0088] Furthermore, when the model training module 260 trains the reinforcement learning model based on the state space input using the advantage-weighted actor-critic algorithm to obtain the trained reinforcement learning model, the model training module 260 is specifically used for: The state space input is used as the input to the actor network, and the action vector is output according to the preset action rules of the actor network; the action vector includes the change of the proportional coefficient, the change of the integral coefficient, the change of the differential coefficient, and the target temperature offset. The state space input and the action vector are used together as the input to the critic network, and a value assessment is output according to the preset reward function of the critic network. The actor network is adjusted according to the value assessment, and the reinforcement learning model is iteratively optimized until the model converges, thus obtaining the trained reinforcement learning model.

[0089] Furthermore, when the heating control module 230 adjusts the output duty cycle of the proportional-integral-derivative controller based on the parameter change and the target temperature offset to adjust the heating power of the heating film of the target solid-state battery, the heating control module 230 is specifically used for: The proportional coefficient, integral coefficient, and derivative coefficient of the proportional-integral-derivative controller are updated based on the changes in the parameters; the changes in the parameters include changes in the proportional coefficient, integral coefficient, and derivative coefficient. The temperature setpoint is updated based on the target temperature offset; wherein the updated temperature setpoint is the sum of the initial target temperature and the target temperature offset; Based on the updated proportional coefficient, integral coefficient, derivative coefficient, and temperature setpoint, as well as the real-time battery temperature, the duty cycle of the proportional-integral-derivative controller is calculated. The duty cycle of the proportional-integral-derivative controller is output to the heating film drive circuit of the target solid-state battery to adjust the heating power of the heating film.

[0090] This application provides a thermal control device for a solid-state battery, comprising: a data acquisition module for acquiring real-time state data of a target solid-state battery; a reinforcement learning module for inputting the real-time state data into a trained reinforcement learning model to obtain the parameter changes of a proportional-integral-derivative (PID) controller and the target temperature offset of the target solid-state battery; and a heating control module for adjusting the output duty cycle of the PID controller based on the parameter changes and the target temperature offset to adjust the heating power of the heating film of the target solid-state battery. In this way, by dynamically optimizing the control parameters of the PID controller in real time through the trained reinforcement learning model, the control accuracy of the solid-state battery thermal control is ensured while reducing the response time and improving the control efficiency.

[0091] Based on the same application concept, please refer to Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 4 As shown, the electronic device 400 includes a processor 410, a memory 420, and a bus 430.

[0092] The memory 420 stores machine-readable instructions executable by the processor 410. When the electronic device 400 is running, the processor 410 and the memory 420 communicate through the bus 430. When the machine-readable instructions are executed by the processor 410, they perform the steps of the thermal control method for solid-state batteries provided in the above embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0093] Based on the same concept, this application also provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it executes the steps of the thermal control method for solid-state batteries provided in the above embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0094] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described apparatus and unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0095] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0096] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0097] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0098] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0099] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0100] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A thermal control method for a solid-state battery, characterized in that, The method includes: Acquire real-time status data of the target solid-state battery; The real-time state data is input into the trained reinforcement learning model to obtain the parameter changes of the proportional-integral-derivative controller and the target temperature offset of the target solid-state battery. Based on the parameter changes and the target temperature offset, the output duty cycle of the proportional-integral-derivative controller is adjusted to adjust the heating power of the heating film of the target solid-state battery.

2. The thermal control method for solid-state batteries according to claim 1, characterized in that, Before acquiring real-time state data of the target solid-state battery, the reinforcement learning model is trained according to the following steps: A solid-state battery thermal simulation model was established, and multi-condition simulation experiments were conducted based on the solid-state battery thermal simulation model to obtain multi-condition simulation generation data. Based on the multi-condition simulation data and the historical operating data of the solid-state battery, the state space input of the reinforcement learning model is determined. Based on the state space input, the reinforcement learning model is trained using the advantage-weighted actor-critic algorithm to obtain the trained reinforcement learning model.

3. The thermal control method for solid-state batteries according to claim 2, characterized in that, The state space input of the reinforcement learning model includes battery state, environmental state, target state and constraints; the battery state includes state of charge, battery temperature, battery temperature error and charge / discharge rate; the environmental state includes coolant temperature, ambient temperature and coolant flow rate; the target state and constraints include desired temperature, maximum allowable temperature rise rate and safe temperature threshold.

4. The thermal control method for solid-state batteries according to claim 3, characterized in that, The step of training the reinforcement learning model based on the state space input using the advantage-weighted actor-critic algorithm to obtain the trained reinforcement learning model includes: The state space input is used as the input to the actor network, and the action vector is output according to the preset action rules of the actor network; the action vector includes the change of the proportional coefficient, the change of the integral coefficient, the change of the differential coefficient, and the target temperature offset. The state space input and the action vector are used together as the input to the critic network, and a value assessment is output according to the preset reward function of the critic network. The actor network is adjusted according to the value assessment, and the reinforcement learning model is iteratively optimized until the model converges, thus obtaining the trained reinforcement learning model.

5. The thermal control method for solid-state batteries according to claim 4, characterized in that, The preset action rules include: Based on the range of the absolute value of the battery temperature error, set the corresponding changes in the proportional coefficient, integral coefficient, and derivative coefficient; Set the corresponding target temperature offset based on the range of the state of charge.

6. The thermal control method for a solid-state battery according to claim 4, characterized in that, The preset reward function includes penalty terms corresponding to temperature error constraints, energy consumption constraints, temperature constraints, temperature rise rate constraints, charge / discharge rate constraints, and heat dissipation efficiency constraints.

7. The thermal control method for solid-state batteries according to claim 1, characterized in that, The step of adjusting the output duty cycle of the proportional-integral-derivative controller based on the parameter change and the target temperature offset to adjust the heating power of the heating film of the target solid-state battery includes: The proportional coefficient, integral coefficient, and derivative coefficient of the proportional-integral-derivative controller are updated based on the changes in the parameters; the changes in the parameters include changes in the proportional coefficient, integral coefficient, and derivative coefficient. The temperature setpoint is updated based on the target temperature offset; wherein the updated temperature setpoint is the sum of the initial target temperature and the target temperature offset; Based on the updated proportional coefficient, integral coefficient, derivative coefficient, and temperature setpoint, as well as the real-time battery temperature, the duty cycle of the proportional-integral-derivative controller is calculated. The duty cycle of the proportional-integral-derivative controller is output to the heating film drive circuit of the target solid-state battery to adjust the heating power of the heating film.

8. A thermal control device for a solid-state battery, characterized in that, The thermal control device for the solid-state battery includes: The data acquisition module is used to acquire real-time status data of the target solid-state battery; The reinforcement learning module is used to input the real-time state data into the trained reinforcement learning model to obtain the parameter changes of the proportional-integral-derivative controller and the target temperature offset of the target solid-state battery. The heating control module is used to adjust the output duty cycle of the proportional-integral-derivative controller based on the parameter change and the target temperature offset, so as to adjust the heating power of the heating film of the target solid-state battery.

9. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor communicates with the memory via the bus. The machine-readable instructions are executed by the processor to perform the steps of the thermal control method for a solid-state battery as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the thermal control method for a solid-state battery as described in any one of claims 1 to 7.

Images