A dynamic output control method and system based on real-time feedback of battery boost

Abstract

The present invention belongs to the field of battery output control, and in particular relates to a dynamic output control method and system based on real-time feedback of battery boost. The method constructs a battery pack output control subsystem, integrates target voltage prediction and multi-level servo control models, accurately predicts and optimizes the target voltage point and boost rate of the battery pack in different health states, and monitors the temperature change and voltage fluctuation during the battery boost process in real time. A bidirectional DC / DC converter is used to achieve fine PWM boost control. At the same time, through a dynamic feedback mechanism, the target voltage point and boost rate are continuously adjusted to ensure that both temperature control changes and voltage target point fluctuations tend to zero, thereby achieving efficient and stable boost output of the battery pack and significantly improving the overall performance and service life of the battery system.

Description

Technical Field

[0001] The present invention belongs to the field of battery output control, and in particular relates to a dynamic output control method and system based on real-time feedback of battery boost. Background Art

[0002] With the development of technology, although lithium iron phosphate batteries (LFP) are now widely used, the design of the inverter still follows this standard, and different energy storage needs are met by adjusting the series and parallel connection mode of the battery pack and the cell capacity; the existing technology provides two implementation methods: one is to use 50Ah cells and adopt the 2P16S method (that is, two cells are connected in parallel and then connected in series 16 times); the other is to use 100Ah cells and adopt the 1P16S method (that is, 16 cells are directly connected in series); the 50Ah solution is favored because of its low cost and high flexibility, however, it requires more cells to be connected in parallel to achieve large-capacity energy storage, which not only increases the complexity of the system, but may also introduce stability problems.

[0003] For example, a Chinese patent with publication number CN112636438A discloses a boost-type balanced discharge circuit and control method for two series-connected battery packs, including two series-connected battery packs, a boost inductor, a boost circuit, an output filter circuit, an output voltage controller, a battery pack balanced discharge controller and a PWM generator; the two series-connected battery packs, the boost inductor, the boost circuit and the output filter circuit are electrically connected in sequence for storage, boosting and filtering, and are controlled by the output voltage controller, the battery pack balanced discharge controller and the PWM generator.

[0004] For example, a Chinese patent with authorization announcement number CN115085314B discloses a battery asynchronous boost charging control circuit, chip, system and method, including: a boost control module, which controls the boost power tube in the boost circuit; controls the input voltage boost to obtain a boost signal in the trickle charging mode and the constant current charging mode, and controls the charging current to be a first set current in the constant current charging mode; controls to stop boosting in the constant voltage charging mode; a charging module, which charges the battery based on the boost signal in the trickle charging mode and the constant current charging mode, and controls the charging current to be a second set current in the trickle charging mode; and a battery balancing control module, which adjusts the charging currents of the first and second batteries respectively based on the relative voltages of the first and second batteries, thereby achieving voltage balancing.

[0005] The above-mentioned prior art has the following problems: the prior art lacks accurate calculation and control of the battery pack boost parameters and the target boost point, resulting in large energy loss during the battery pack boost process and difficulty in maintaining a stable state during the boost process. In order to solve the above problems, the present invention provides a dynamic output control method and system based on real-time feedback of battery boost. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a dynamic output control method and system based on real-time feedback of battery boost. The method constructs a battery pack output control subsystem, integrates target voltage prediction and multi-level servo control models, accurately predicts and optimizes the target voltage point and boost rate of the battery pack under different health states, and monitors the temperature changes and voltage fluctuations in the battery boost process in real time. A bidirectional DC / DC converter is used to achieve fine PWM boost control. At the same time, through a dynamic feedback mechanism, the target voltage point and boost rate are continuously adjusted to ensure that both temperature control changes and voltage target point fluctuations tend to zero, thereby achieving efficient and stable boost output of the battery pack and significantly improving the overall performance and service life of the battery system.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A dynamic output control method based on real-time feedback of battery boost, comprising:

[0009] S1: Build a 1P6S battery pack output control subsystem, obtain the battery pack health status and corresponding attribute data and power output data under the health status, and pre-process the acquired data;

[0010] S2: Build and train a boost prediction model, and embed the trained boost prediction model into the output control subsystem. Input the data obtained from the health status of the battery pack into the boost prediction model, output the target voltage point and boost rate under the health status, and use the acquired boost rate to calculate the target temperature control change rate under the corresponding boost rate.

[0011] S3: Build and train a multi-level servo control model, embed the pre-trained multi-level servo control model into the output control subsystem, input the real-time power output data and the predicted target voltage point and boost rate at the current moment into the output control subsystem to output a joint multi-level control instruction;

[0012] S4: input the obtained joint multi-level control instruction into the bidirectional DC / DC converter to generate PWM boost control and state conversion signals, control the output state and boost rate of the battery pack during the boost process, and monitor the real-time temperature change rate and the actual target boost value during the boost process;

[0013] S5: subtract the acquired real-time temperature change rate and the actual target boost value from the target temperature control change rate and the predicted target voltage point corresponding to the current moment, to obtain the temperature control change difference and the voltage target point fluctuation difference, and input the acquired temperature control change difference and the voltage target point fluctuation difference into the boost prediction model to optimize the target voltage point sequence and the boost rate sequence, so that the temperature control change difference and the voltage target point fluctuation difference are both equal to 0;

[0014] The specific steps of calculating the target temperature control change rate at the corresponding boost rate in S2 include:

[0015] S211, based on the predicted boost rate, target voltage point, and initial voltage and the resistance inside the battery pack , calculate the corresponding boost time length T and the corresponding output current value at each time point;

[0016] S212, calculate the heat loss of the battery pack during the boost process based on the calculated output current value and the internal resistance of the battery pack at each time point, and calculate the target temperature control change rate at the boost rate corresponding to the kth health state based on the calculated heat loss and battery pack attribute data using the laws of thermodynamics , the specific formula is:

[0017] ,

[0018] Where h represents the air convection heat transfer coefficient, A represents the heat exchange area on the battery pack surface, represents the average temperature of the ambient air where the battery pack is located at time t, m represents the mass of the battery, Represents the average specific heat capacity of battery materials; represents the voltage boost rate of the battery pack at time t in the kth health state, represents the battery pack temperature at time t in the kth health state;

[0019] The multi-level servo control model includes a control adjustment layer and a command output layer; the specific steps of constructing the control adjustment layer include:

[0020] S301: construct the current input state according to the acquired battery pack attribute data, power output data, target voltage point and boost rate data ,in, Indicates the state of charge of the battery pack at time t, represents the evaluation score interval corresponding to the evaluation level of the battery pack in the kth health state, represents the battery pack health status assessment score in the kth health state, represents the voltage value of the battery pack at time t in the kth health state, Indicates the target voltage point of the battery pack in the kth health state, Indicates the limited remaining state of charge corresponding to the battery pack in the kth health state;

[0021] S302, setting multi-level trigger information according to the current input state of S301, specifically including:

[0022] When the battery pack meets the When the target voltage point and Execute boost information, when the battery pack meets the When the battery pack health status is adjusted, the new target voltage point and information;

[0023] When triggered according to After executing the boost information, if the boost process does not meet , or , then trigger the information of adjusting the boost rate in parallel; and represents the absolute value of the target voltage point error value and the corresponding threshold value under the kth healthy state corresponding to time t, and represents the absolute value of the boost rate error value under the kth healthy state corresponding to time t and the corresponding threshold; Indicates the real-time change of battery pack temperature with boost rate per unit time in the kth health state; It represents the target temperature control change rate at the corresponding boost rate at time t in the kth health state;

[0024] When the information of adjusting the boost rate is triggered, if the battery pack meets the When the condition is met, the information of converting the battery pack output state is triggered. When the battery is disconnected from the output state, otherwise it will automatically connect to the battery pack for power output;

[0025] The specific steps to control the construction of adjustment layers also include:

[0026] S303, constructing the current execution action based on the set multi-level trigger information:

[0027] ,

[0028] in, Indicates triggering according to The boost action corresponding to the boost information is executed. Indicates triggering the adjustment of the current battery pack health status to obtain a new target voltage point and The execution action, Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that an action is being performed , or During the process, the battery pack meets The information corresponds to the action of converting the output state of the battery pack executed;

[0029] S304: Construct a reward function based on the execution action of S303 , specifically:

[0030] ,

[0031] in, It represents the reward for adjusting the boost rate for the gth time in the kth health state so that the boost process meets the triggering condition of the corresponding execution action. Indicates correct execution The corresponding rewards, , and is the gain coefficient of the corresponding reward;

[0032] The specific steps of constructing the instruction output layer include:

[0033] S307, construct the command output layer based on the control algorithm, and obtain and Initialize the control parameters in the instruction output layer, and use the initialized parameters and the obtained , and output control adjustment to generate joint multi-level control instructions , the joint multi-level control instructions are as follows:

[0034] ,

[0035] in, and is the joint control factor, when Execute the corresponding action , Execute the corresponding action , In turn, it means to perform actions by controlling the adjustment layer. , , The corresponding evaluation value generated at Corresponding execution , Do not execute ; for The zth evaluation value in , represents the proportional coefficient within the parameter in the instruction output layer, Represents the integral coefficient within the parameter in the instruction output layer, Represents the differential coefficient within the parameter in the instruction output layer.

[0036] Specifically, the battery pack health status includes K different levels of health status; the specific steps of building and training the boost prediction model include:

[0037] S201, obtaining battery pack historical attribute data, power output data, target voltage point and boost rate data to construct input sequence data , where X represents the battery pack attribute data and power output data under different health states of the battery pack, and Y represents the predicted target data, i.e., the historical target voltage point and boost rate data;

[0038] S202, construct a prediction sub-model in the boost prediction model based on the BP neural network, and input the sequence data Input into the prediction sub-model to predict the target voltage point and boost rate data under K healthy states;

[0039] S203: Setting the error threshold under the kth health state , the target voltage point and boost rate data under the kth healthy state are obtained by prediction, and the target error data is calculated ,in and represents the absolute value of the target voltage point error value and the corresponding threshold value under the kth healthy state corresponding to time t, and represents the absolute value of the boost rate error value under the kth healthy state corresponding to time t and the corresponding threshold;

[0040] S204, when and , the target voltage point and boost rate value that meet the conditions are obtained.

[0041] Specifically, the specific steps of building and training the boost prediction model also include:

[0042] S205: Target error data obtained according to S203 , BP neural network weights and error thresholds, randomly generate N initial gene populations of length P ;in, Represents the i-th weight among the n neural network weights of the prediction sub-model;

[0043] S206, using the acquired power output data and the initial gene population target error to construct the fitness function and constraint conditions corresponding to the target voltage point and boost rate at the current time t, specifically including:

[0044] ,

[0045] , , , ,

[0046] Where T represents the time length for the battery pack to complete a voltage boost. It represents the real-time change of battery pack temperature with boost rate per unit time in the kth health state. represents the battery pack temperature at time t in the kth health state, represents the voltage boost rate of the battery pack at time t in the kth health state, represents the discharge depth of the battery pack at time t in the kth health state, It represents the target temperature control change rate at the corresponding boost rate at time t in the kth health state, Indicates the rated state of charge of the battery pack. Indicates the limited remaining state of charge corresponding to the battery pack in the kth health state;

[0047] S207, set the maximum number of iterations G and fitness threshold , iterate the genetic algorithm and compare the fitness function value after each iteration. If the fitness function value corresponding to the jth gene population is continuously in the iteration If all the genes are in increasing state, the jth population is discarded, otherwise the jth gene population is retained;

[0048] S208, repeat the process of S207 for G iterations, and obtain the minimum fitness function value during the iteration and the corresponding fitness function value is less than The n neural network weights iterated in the corresponding gene population are fed back to the prediction sub-model to optimize the prediction process in real time and obtain the trained boost prediction model.

[0049] A dynamic output control system based on real-time feedback of battery boost, comprising: a data processing module, a boost control module, an output control module and a feedback optimization module;

[0050] A data processing module, used to obtain the health status of the battery pack and the attribute data under the corresponding health status, power output data, target voltage point and boost rate data, and pre-process the obtained data;

[0051] The boost control module includes a boost prediction unit and a multi-stage servo control unit;

[0052] A boost prediction unit is used to predict the target voltage point and boost rate corresponding to different health states of the battery pack through a constructed boost prediction model based on the data obtained from the corresponding battery health state;

[0053] The multi-level servo control unit is used to output a joint multi-level control instruction through a multi-level servo control model according to the acquired real-time power output data and the predicted target voltage point and boost rate.

[0054] Specifically, the output control module includes a signal generating unit and a state monitoring unit; the signal generating unit is used to generate a PWM signal according to the acquired multi-stage boost control instruction to control the working state of the bidirectional DC / DC converter;

[0055] A state monitoring unit, used for real-time monitoring of the temperature change and actual target boost value of the battery pack under the control of the PWM signal generated by the multi-stage boost control instruction;

[0056] The feedback optimization module includes a difference calculation unit;

[0057] The difference calculation unit is used to calculate the difference between the actual temperature change rate and the target temperature control change rate, and the difference between the actual target boost value and the predicted target voltage point.

[0058] A computer-readable storage medium stores computer instructions, which, when executed, execute a dynamic output control method based on real-time feedback of battery boost.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] In view of the deficiencies of the prior art, the present invention can dynamically adjust the boost strategy according to the actual health status of the battery pack through a built-in boost prediction model, achieve the optimal match between the target voltage point and the boost rate, and then calculate the corresponding temperature control change rate, effectively avoiding the overheating problem; in addition, a multi-level servo control model is adopted, combined with real-time power output data and prediction results, to achieve refined management of the output state, ensuring the maximization of energy efficiency during the battery pack boost process; finally, through continuous optimization of the boost prediction model, the temperature control change difference and the voltage target point fluctuation difference during the boost process are made close to zero, which greatly reduces energy loss and ensures the smooth operation of the boost process;

[0061] The present invention uses a BP neural network combined with a genetic algorithm optimization method to finely control the boost process of the battery pack at different health levels, effectively reducing the energy loss of the battery pack during the boost process due to target voltage point fluctuations and temperature control changes, and significantly improving the overall performance and service life of the battery pack. In addition, by real-time monitoring of the actual temperature control change rate and target boost value of the battery pack during the boost process, and comparing and analyzing them with the predicted value, the boost strategy can be adjusted in a timely manner, further enhancing the adaptability and stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of a dynamic output control method based on real-time feedback of battery boost in embodiment 1 of the present invention;

[0063] Figure 2 This is a diagram of the boost prediction model architecture of Example 2 of the present invention;

[0064] Figure 3 This is a module diagram of a dynamic output control system based on real-time feedback of battery boost according to Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0065] Example 1

[0066] See also Figure 1 , an embodiment provided by the present invention: a dynamic output control method based on real-time feedback of battery boost, comprising the steps of:

[0067] S1: construct a 1P6S battery pack output control subsystem, obtain the battery pack health status and attribute data and power output data under the corresponding health status, and pre-process the acquired data; further, the battery pack output control subsystem in this embodiment includes a battery pack composed of 1 parallel and 6 series, a bidirectional DC / DC and other electrical equipment, such as load equipment; the battery pack is composed of 280Ah battery cells; the battery pack attribute data includes voltage, current, SOC, SOH, SOF, temperature change rate per unit time, current service life and remaining service life;

[0068] Furthermore, in this embodiment, the specific steps of obtaining the health status of the battery pack include:

[0069] S101, constructing a health assessment model through a comprehensive assessment algorithm, constructing an assessment factor based on the historical attribute data and power output data of the battery pack under different health states, and constructing K assessment indicators and corresponding scoring intervals through expert experience method; further, the K assessment indicators here are specifically set by those skilled in the art according to the actual working state of the battery pack;

[0070] S102, inputting the constructed evaluation factors, evaluation indicators and corresponding scoring intervals into the health assessment model for training, and obtaining a trained health assessment model;

[0071] S103, deploying the trained health assessment model to the battery pack output control subsystem, and calculating the health status score and health status level of the battery pack at each time point according to the real-time acquired battery pack attribute data and power output data;

[0072] S2: Build and train a boost prediction model, and embed the trained boost prediction model into the output control subsystem. Input the data obtained from the health status of the battery pack into the boost prediction model, output the target voltage point and boost rate under the health status, and use the acquired boost rate to calculate the target temperature control change rate under the corresponding boost rate.

[0073] See also Figure 2 Furthermore, the specific steps of the boost prediction model in this embodiment include:

[0074] S201, obtaining battery pack historical attribute data, power output data, target voltage point and boost rate data to construct input sequence data , where X represents the battery pack attribute data and power output data under different health states of the battery pack, and Y represents the predicted target data, i.e., the historical target voltage point and boost rate data;

[0075] S202, construct a prediction sub-model in the boost prediction model based on the BP neural network, and input the sequence data Input into the prediction sub-model to predict the target voltage point and boost rate data under K health states; further, the K health states in this step are obtained by the health assessment model based on the acquired battery pack attribute data and power output data;

[0076] S203: Setting the error threshold under the kth health state , the target voltage point and boost rate data under the kth healthy state are obtained by prediction, and the target error data is calculated ,in and represents the absolute value of the target voltage point error value and the corresponding threshold value under the kth healthy state corresponding to time t, and represents the absolute value of the boost rate error value under the kth healthy state corresponding to time t and the corresponding threshold value; further in this embodiment, it is specifically set All are 0, and those skilled in the art can also make specific settings according to specific accuracy requirements;

[0077] S204, when and , the target voltage point and boost rate value that meet the conditions are obtained.

[0078] S205: Target error data obtained according to S203 , BP neural network weights and error thresholds, randomly generate N initial gene populations of length P ;in, Represents the i-th weight among the n neural network weights of the prediction sub-model, i=1,...,n;

[0079] S206, using the acquired power output data and the initial gene population target error to construct the fitness function and constraint conditions corresponding to the target voltage point and boost rate at the current time t, specifically including:

[0080] ,

[0081] , , , ,

[0082] Where T represents the time length for the battery pack to complete a voltage boost. It represents the real-time change of battery pack temperature with boost rate per unit time in the kth health state. represents the battery pack temperature at time t in the kth health state, represents the voltage boost rate of the battery pack at time t in the kth health state, represents the discharge depth of the battery pack at time t in the kth health state, It represents the target temperature control change rate at the corresponding boost rate at time t in the kth health state, Indicates the rated state of charge of the battery pack. Indicates the limited remaining state of charge corresponding to the battery pack in the kth health state; further, in this embodiment, the first and second constraints are to limit the error of the prediction result, which is used to improve the accuracy of the prediction. The third constraint corresponds to the relationship between the boost rate and the temperature change. This constraint ensures that the real-time temperature change rate of the battery pack is always less than or equal to the target temperature control change rate within the time length of a cycle, that is, the change rate of the battery pack temperature with the boost rate is always less than or equal to the change rate of heat when the battery pack reaches the maximum temperature for normal operation. The fourth constraint indicates that the amount of electricity released by the battery pack is always less than the maximum amount of electricity allowed to be released by the battery pack at the current moment, that is, excluding the released electricity, the remaining electricity can ensure that the current health state of the battery pack is in the best state; further, in this embodiment, The data are obtained by technicians in this field through real-time monitoring by sensors;

[0083] S207, setting the maximum number of iterations G and fitness threshold of the genetic algorithm , iterate the genetic algorithm and compare the fitness function value after each iteration. If the fitness function value corresponding to the jth gene population is continuously in the iteration If all the genes are in increasing state, the jth population is discarded, otherwise the jth gene population is retained;

[0084] S208, repeat the process of S207 for G iterations, and obtain the minimum fitness function value during the iteration and the corresponding fitness function value is less than The n neural network weights iterated in the corresponding gene population are fed back to the prediction sub-model, the prediction process is optimized in real time, and the trained boost prediction model is obtained;

[0085] This process uses the BP neural network as the prediction sub-model and combines it with the genetic algorithm to optimize the weights of the neural network, so that the model can more accurately predict the target voltage point and boost rate under different health states. This allows the system to maintain a high prediction accuracy even if the battery performance changes over time, thereby improving the reliability and stability of the overall system. In addition, by introducing the temperature control change rate constraint related to the boost rate, it ensures that the temperature of the battery during the boost process will not exceed the safe range, effectively avoiding safety problems that may be caused by excessive temperature; finally, by optimizing the model in real time, it can adapt to changes in battery performance over time, thereby ensuring the long-term effectiveness of the prediction model, which not only helps to extend the battery life, but also improves the energy efficiency of the entire system.

[0086] Furthermore, in this embodiment, the specific steps of calculating the target temperature control change rate at the corresponding boost rate include:

[0087] S211, based on the predicted boost rate, target voltage point, and initial voltage and the resistance inside the battery pack , calculate the corresponding boost time length T and the corresponding output current value at each time point;

[0088] S212, calculate the heat loss of the battery pack during the boost process based on the calculated output current value and the internal resistance of the battery pack at each time point, and calculate the target temperature control change rate at the boost rate corresponding to the kth health state based on the calculated heat loss and battery pack attribute data using the laws of thermodynamics , the specific formula is:

[0089] ,

[0090] Where h represents the air convection heat transfer coefficient, A represents the heat exchange area on the battery pack surface, represents the average temperature of the ambient air where the battery pack is located at time t, m represents the mass of the battery, represents the average specific heat capacity of the battery material; further, in the above formula, The calculation was obtained in The current change at each moment during the boost process is the boost rate. At the same time, the corresponding battery energy loss in a complete boost process is calculated by the work formula, and the heat dissipated into the air by the battery during the boost process is calculated by the Stefan-Boltzmann law. The battery energy loss minus the heat dissipated into the air is the remaining battery current accumulated heat, and then the second law of thermodynamics is used to calculate the The temperature change value corresponding to a complete boost process of the boost rate is obtained by dividing the obtained temperature change value by T. The corresponding temperature change rate is used as The temperature change rate of the battery during the boosting process is limited by the temperature control change rate during the boosting process, which not only accurately controls the parameters during the boosting process, but also further increases the service life of the battery.

[0091] S3: Build and train a multi-level servo control model, embed the pre-trained multi-level servo control model into the output control subsystem, input the real-time power output data and the predicted target voltage point and boost rate at the current moment into the output control subsystem to output a joint multi-level control instruction;

[0092] Furthermore, in this embodiment, the multi-level servo control model includes a control adjustment layer and a command output layer; the specific steps of constructing the control adjustment layer include:

[0093] S301: construct the current input state according to the acquired battery pack attribute data, power output data, target voltage point and boost rate data ,in, Indicates the state of charge of the battery pack at time t, represents the evaluation score interval corresponding to the evaluation level of the battery pack in the kth health state, represents the battery pack health status assessment score in the kth health state, represents the voltage value of the battery pack at time t in the kth health state, represents the target voltage point corresponding to the battery pack in the kth healthy state; further, in this embodiment, the target voltage of the battery pack for initial use in this embodiment is set to 51.2V. When the battery changes due to the change of service life, the setting is corrected in real time according to the result predicted by the above-mentioned boost prediction model, so that each boost can be kept within the safe range of the battery;

[0094] S302, setting multi-level trigger information according to the current input state of S301, specifically including:

[0095] When the battery pack meets the When the target voltage point and Execute boost information, when the battery pack meets the When the battery pack health status is adjusted, the new target voltage point and information;

[0096] When triggered according to After executing the boost information, if the boost process does not meet , or , then trigger the information of adjusting the boost rate in parallel;

[0097] When the information of adjusting the boost rate is triggered, if the battery pack meets the When the condition is met, the information of converting the battery pack output state is triggered. The battery output state is disconnected when the battery is in the off state, otherwise the battery pack is automatically connected for power output.

[0098] S303, constructing the current execution action based on the set multi-level trigger information:

[0099] ,

[0100] in, Indicates triggering according to The boost action corresponding to the boost information is executed. Indicates triggering the adjustment of the current battery pack health status to obtain a new target voltage point and The execution action, Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that an action is being performed , or During the process, the battery pack meets The information corresponds to the action of converting the output state of the battery pack executed;

[0101] S304: Construct a reward function based on the execution action of S303 , specifically:

[0102] ,

[0103] in, It represents the reward for adjusting the boost rate for the gth time in the kth health state so that the boost process meets the triggering condition of the corresponding execution action. Indicates correct execution The corresponding rewards, , and is the gain coefficient of the corresponding reward;

[0104] S305, constructing the control adjustment layer based on reinforcement learning, and inputting the constructed current input state, multi-level trigger information, current execution action and reward function into the control adjustment layer for training, and obtaining the trained control adjustment layer;

[0105] S306, inputting the real-time acquired battery pack attribute data, power output data, target voltage point and boost rate into the control adjustment layer to obtain the output control adjustment amount;

[0106] S307, construct the command output layer based on the control algorithm, and obtain and Initialize the control parameters in the instruction output layer, and use the initialized parameters and the obtained , and output control adjustment to generate joint multi-level control instructions , the joint multi-level control instructions are as follows:

[0107]

[0108] in, and is the joint control factor, when Execute the corresponding action , Execute the corresponding action , In turn, it means to perform actions by controlling the adjustment layer. , , The corresponding evaluation value generated at Corresponding execution , Do not execute ; for The zth evaluation value in . That is, when the corresponding execution , , Which action in The evaluation value corresponding to the execution action; represents the proportional coefficient within the parameter in the instruction output layer, Represents the integral coefficient within the parameter in the instruction output layer, represents the differential coefficient of the parameter in the instruction output layer; further, in this embodiment, for the combined multi-level control instruction The specific generation execution method is: first pass The specific assignment determines the execution action still , when the assignment is completed or After that, the corresponding lower-level action is executed , or , Corresponding execution action The execution level and and The same and independent of each other;

[0109] S308, integrating the constructed control adjustment layer and instruction output layer obtained above into the multi-level servo control model, and building the multi-level servo control model into the battery pack output control subsystem, acquiring the battery pack power output data and the target voltage point and boost rate obtained by the corresponding health status prediction in real time, and outputting the corresponding joint multi-level control instruction through the multi-level servo control model built into the battery pack output control subsystem;

[0110] The control adjustment layer in this process can dynamically adjust the boost strategy according to the actual state and health status of the battery pack to ensure efficient and stable operation of the battery pack under different working conditions; in addition, by setting multi-level trigger information, it can promptly detect and correct abnormal conditions during the battery pack boost process, effectively avoiding energy loss or equipment damage caused by overvoltage or undervoltage; the training method based on reinforcement learning further improves the adaptability and robustness of the control strategy, enabling the model to make optimal decisions in a complex operating environment; the command output layer is responsible for converting the adjusted control signal into actual control commands to ensure the continuity and controllability of the battery pack output.

[0111] S4: input the obtained joint multi-level control instruction to the bidirectional DC / DC converter to generate PWM boost control and state conversion signals, control the output state and boost rate of the battery pack during the boost process, and monitor the real-time temperature change rate and the actual target boost value during the boost process; further, the output state in this embodiment refers to whether the battery pack is connected or disconnected at the current moment;

[0112] S5: Subtract the acquired real-time temperature change rate and the actual target boost value from the target temperature control change rate and the predicted target voltage point corresponding to the current moment to obtain the temperature control change difference and the voltage target point fluctuation difference, and input the acquired temperature control change difference and voltage target point fluctuation difference into the boost prediction model to optimize the target voltage point sequence and the boost rate sequence, so that the temperature control change difference and the voltage target point fluctuation difference are both equal to 0.

[0113] Example 2

[0114] See also Figure 3 , another embodiment provided by the present invention: a dynamic output control system based on real-time feedback of battery boost, comprising: a data processing module, a boost control module, an output control module and a feedback optimization module;

[0115] The data processing module is used to obtain the health status of the battery pack and the attribute data under the corresponding health status, power output data, target voltage point and boost rate data, and pre-process the obtained data;

[0116] The boost control module is used to predict the target voltage point and boost rate in real time and output control instructions according to these predicted values; the boost control module includes a boost prediction unit and a multi-stage servo control unit;

[0117] The boost prediction unit is used to predict the target voltage point and boost rate corresponding to different health states of the battery pack through the constructed boost prediction model according to the data obtained from the corresponding battery health state;

[0118] The multi-level servo control unit is used to output a joint multi-level control instruction through a multi-level servo control model according to the acquired real-time power output data and the predicted target voltage point and boost rate;

[0119] The output control module is used to generate a PWM control signal to control the boost process of the battery pack and monitor the state change of the battery in real time during the boost process; the output control module includes a signal generating unit and a state monitoring unit; the signal generating unit is used to generate a PWM signal according to the acquired multi-stage boost control instruction to control the working state of the bidirectional DC / DC converter;

[0120] The state monitoring unit is used to monitor the real-time temperature change and actual target boost value of the battery pack under the control of the PWM signal generated by the multi-stage boost control instruction;

[0121] The feedback optimization module is used to calculate the difference between the actual measured value and the predicted value to optimize the prediction model and the prediction result; the feedback optimization module includes a difference calculation unit and a model optimization unit;

[0122] The difference calculation unit is used to calculate the difference between the actual temperature change rate and the target temperature control change rate, and the difference between the actual target boost value and the predicted target voltage point;

[0123] The model optimization unit is used to feed back the difference to the boost prediction model to adjust the model parameters and optimize the prediction performance, while further improving the accuracy of the control instructions output by the multi-level servo control model.

[0124] Example 3

[0125] A computer-readable storage medium stores computer instructions, which, when executed, execute a dynamic output control method based on real-time feedback of battery boost.

[0126] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements a dynamic output control method based on real-time feedback of battery boost when executing the computer program.

[0127] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the scope of protection of the purpose of the present invention and the claims, and all of these are within the protection of the present invention.

Claims

1. A dynamic output control method based on real-time feedback of battery boost, characterized in that: include: S1: Build a 1P6S battery pack output control subsystem, obtain the battery pack health status and corresponding attribute data and power output data under the health status, and pre-process the acquired data; S2: Build and train a boost prediction model, and embed the trained boost prediction model into the output control subsystem. Input the data obtained from the health status of the battery pack into the boost prediction model, output the target voltage point and boost rate under the health status, and use the acquired boost rate to calculate the target temperature control change rate under the corresponding boost rate. S3: Build and train a multi-level servo control model, embed the pre-trained multi-level servo control model into the output control subsystem, input the real-time power output data and the predicted target voltage point and boost rate at the current moment into the output control subsystem to output a joint multi-level control instruction; S4: input the obtained joint multi-level control instruction into the bidirectional DC / DC converter to generate PWM boost control and state conversion signals, control the output state and boost rate of the battery pack during the boost process, and monitor the real-time temperature change rate and the actual target boost value during the boost process; S5: subtract the acquired real-time temperature change rate and the actual target boost value from the target temperature control change rate and the predicted target voltage point corresponding to the current moment, to obtain the temperature control change difference and the voltage target point fluctuation difference, and input the acquired temperature control change difference and the voltage target point fluctuation difference into the boost prediction model to optimize the target voltage point sequence and the boost rate sequence, so that the temperature control change difference and the voltage target point fluctuation difference are both equal to 0; The specific steps of calculating the target temperature control change rate corresponding to the boost rate in S2 include: S211, based on the predicted boost rate, target voltage point, and initial voltage and the resistance inside the battery pack , calculate the corresponding boost time length T and the corresponding output current value at each time point; S212, calculate the heat loss of the battery pack during the boost process based on the calculated output current value and the internal resistance of the battery pack at each time point, and calculate the target temperature control change rate at the boost rate corresponding to the kth health state based on the calculated heat loss and battery pack attribute data using the laws of thermodynamics , the specific formula is: ， Where h represents the air convection heat transfer coefficient, A represents the heat exchange area on the battery pack surface, represents the average temperature of the ambient air where the battery pack is located at time t, m represents the mass of the battery, Represents the average specific heat capacity of battery materials; represents the voltage boost rate of the battery pack at time t in the kth health state, represents the battery pack temperature at time t in the kth health state; The multi-level servo control model includes a control adjustment layer and a command output layer; the specific steps of constructing the control adjustment layer include: S301: construct the current input state according to the acquired battery pack attribute data, power output data, target voltage point and boost rate data ,in, Indicates the state of charge of the battery pack at time t, represents the evaluation score interval corresponding to the evaluation level of the battery pack in the kth health state, represents the battery pack health status assessment score in the kth health state, represents the voltage value of the battery pack at time t in the kth health state, Indicates the target voltage point of the battery pack in the kth health state, Indicates the limited remaining state of charge corresponding to the battery pack in the kth health state; S302, setting multi-level trigger information according to the current input state of S301, specifically including: When the battery pack meets the When the target voltage point and Execute boost information, when the battery pack meets the When the battery pack health status is adjusted, the new target voltage point and information; When triggered according to After executing the boost information, if the boost process does not meet , or , then trigger the information of adjusting the boost rate in parallel; and represents the absolute value of the target voltage point error value and the corresponding threshold value under the kth healthy state corresponding to time t, and represents the absolute value of the boost rate error value under the kth healthy state corresponding to time t and the corresponding threshold; Indicates the real-time change of battery pack temperature with boost rate per unit time in the kth health state; It represents the target temperature control change rate at the corresponding boost rate at time t in the kth health state; When the information of adjusting the boost rate is triggered, if the battery pack meets the When the condition is met, the information of converting the battery pack output state is triggered. When the battery is disconnected from the output state, otherwise it will automatically connect to the battery pack for power output; The specific steps of controlling the construction of the adjustment layer also include: S303, constructing the current execution action based on the set multi-level trigger information: ， in, Indicates triggering according to The boost action corresponding to the boost information is executed. Indicates triggering the adjustment of the current battery pack health status to obtain a new target voltage point and The execution action, Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that when the voltage is not satisfied during the boost process The corresponding adjustment boost rate is executed when the condition is met. Indicates that an action is being performed , or During the process, the battery pack meets The information corresponds to the action of converting the output state of the battery pack executed; S304: Construct a reward function based on the execution action of S303 , specifically: ， in, It represents the reward for adjusting the boost rate for the gth time in the kth health state so that the boost process meets the triggering condition of the corresponding execution action. Indicates correct execution The corresponding rewards, , and is the gain coefficient of the corresponding reward; The specific steps of constructing the instruction output layer include: S307, construct the command output layer based on the control algorithm, and obtain and Initialize the control parameters in the instruction output layer, and use the initialized parameters and the obtained , and output control adjustment to generate joint multi-level control instructions , the joint multi-level control instructions are as follows: ， in, and is the joint control factor, when Execute the corresponding action , Execute the corresponding action , In turn, it means to perform actions by controlling the adjustment layer. , , The corresponding evaluation value generated at Corresponding execution , Do not execute ; for The zth evaluation value in , represents the proportional coefficient within the parameter in the instruction output layer, Represents the integral coefficient within the parameter in the instruction output layer, Represents the differential coefficient within the parameter in the instruction output layer.

2. A dynamic output control method based on real-time feedback of battery boost as claimed in claim 1, characterized in that: The battery pack health status includes K different levels of health status; the specific steps of building and training the boost prediction model include: S201, obtaining battery pack historical attribute data, power output data, target voltage point and boost rate data to construct input sequence data , where X represents the battery pack attribute data and power output data under different health states of the battery pack, and Y represents the predicted target data, i.e., the historical target voltage point and boost rate data; S202, construct a prediction sub-model in the boost prediction model based on the BP neural network, and input the sequence data Input into the prediction sub-model to predict the target voltage point and boost rate data under K healthy states; S203: Setting the error threshold under the kth health state , the target voltage point and boost rate data under the kth healthy state are obtained by prediction, and the target error data is calculated ; S204, when and , the target voltage point and boost rate value that meet the conditions are obtained.

3. A dynamic output control method based on real-time feedback of battery boost as claimed in claim 2, characterized in that: The specific steps of building and training the boost prediction model also include: S205: Target error data obtained according to S203 , BP neural network weights and error thresholds, randomly generate N initial gene populations of length P ;in, Represents the i-th weight among the n neural network weights of the prediction sub-model; S206, using the acquired power output data and the initial gene population target error to construct the fitness function and constraint conditions corresponding to the target voltage point and boost rate at the current time t, specifically including: ， ， ， ， ， Where T represents the time length for the battery pack to complete a voltage boost. represents the discharge depth of the battery pack at time t in the kth health state, It represents the target temperature control change rate at the corresponding boost rate at time t in the kth health state, Indicates the rated state of charge of the battery pack; S207, set the maximum number of iterations G and fitness threshold , iterate the genetic algorithm and compare the fitness function value after each iteration. If the fitness function value corresponding to the jth gene population is continuously in the iteration If all the genes are in increasing state, the jth population is discarded, otherwise the jth gene population is retained; S208, repeat the process of S207 for G iterations, and obtain the minimum fitness function value during the iteration and the corresponding fitness function value is less than The n neural network weights iterated in the corresponding gene population are fed back to the prediction sub-model to optimize the prediction process in real time and obtain the trained boost prediction model.

4. A dynamic output control system based on real-time feedback of battery boost, which is implemented based on a dynamic output control method based on real-time feedback of battery boost as claimed in any one of claims 1 to 3, characterized in that: include: Data processing module, boost control module, output control module and feedback optimization module; The data processing module is used to obtain the health status of the battery pack and the attribute data under the corresponding health status, power output data, target voltage point and boost rate data, and pre-process the obtained data; The boost control module includes a boost prediction unit and a multi-stage servo control unit; The boost prediction unit is used to predict the target voltage point and boost rate corresponding to different health states of the battery pack through the constructed boost prediction model according to the data obtained from the corresponding battery health state; The multi-level servo control unit is used to output a joint multi-level control instruction through a multi-level servo control model according to the acquired real-time power output data and the predicted target voltage point and boost rate.

5. A dynamic output control system based on real-time feedback of battery boost as claimed in claim 4, characterized in that: The output control module includes a signal generating unit and a state monitoring unit; the signal generating unit is used to generate a PWM signal according to the acquired multi-stage boost control instruction to control the working state of the bidirectional DC / DC converter; The state monitoring unit is used to monitor the real-time temperature change and actual target boost value of the battery pack under the control of the PWM signal generated by the multi-stage boost control instruction; The feedback optimization module includes a difference calculation unit; The difference calculation unit is used to calculate the difference between the actual temperature change rate and the target temperature control change rate, and the difference between the actual target boost value and the predicted target voltage point.

6. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the computer instructions are executed, a dynamic output control method based on real-time feedback of battery boost as described in any one of claims 1 to 3 is executed.

Images