Neural network based fuel cell vehicle integrated thermal management system and method

Abstract

The application discloses a fuel cell vehicle integrated thermal management system and method based on a neural network, collects fuel cell thermal management data, converts various operation processes and parameter contents in a stack system into specific values, displays the structure of the stack system and the converted specific values, simulates output characteristics, efficiency and temperature distribution of the stack system under different working conditions, pre-processes and feature-processes the fuel cell thermal management data collected by a sensor data collection module, trains and updates a neural network control module, and controls the stack system, a radiator, a heat exchanger, a circulating water pump, a cockpit, a battery pack and a heating device according to internal parameters obtained through feature processing and training and updating of the neural network control module. The application realizes temperature control of the fuel cell through the more accurate integrated thermal management system module, and realizes higher energy transmission efficiency and effective power output of the stack.

Description

Technical Field

[0001] This invention belongs to the field of thermal management technology, specifically relating to an integrated thermal management system and method for fuel cell vehicles based on neural networks. Background Technology

[0002] To achieve global carbon neutrality and address energy and environmental issues, countries have set carbon peaking and carbon neutrality targets and formulated corresponding plans. Against this backdrop, global automakers have increased their research and development efforts in fuel cell vehicle technology. However, moving fuel cell vehicle technology from the research stage to actual production still faces several challenges. Besides high manufacturing and operating costs, the independent and decentralized nature of the fuel cell thermal management subsystem leads to high energy consumption, significantly impacting the overall vehicle's economics.

[0003] Existing fuel cell systems suffer from the following technical problems, including insufficient integration between the various subsystems, resulting in the inability to precisely control energy consumption between the subsystems, and low energy conversion efficiency of the fuel cell and its surrounding thermal management system. Summary of the Invention

[0004] The purpose of this invention is to provide an integrated thermal management system and method for fuel cell vehicles based on neural networks, which achieves more precise temperature control of the fuel cell through a more precise integrated thermal management system module, thereby achieving higher energy transfer efficiency and effective power output of the fuel cell stack.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is: an integrated thermal management system for fuel cell vehicles based on neural networks, comprising:

[0006] The sensor data collection module is used to collect fuel cell thermal management data, including the temperature of the fuel cell stack system, the temperature of the cooling system, the temperature of the cabin, the voltage and current values ​​of the battery pack, and the ambient temperature of the fuel cell vehicle, in order to obtain the thermal state of the stack, the performance of the cooling system, the working status of the battery pack, and the current output.

[0007] The fuel cell stack module is used to convert various operating processes and parameter contents within the fuel cell stack system into specific values, display the structure of the fuel cell stack system and the converted specific values, and simulate the output characteristics, efficiency, and temperature distribution of the fuel cell stack system under different operating conditions. The operating process includes the generation process of reversible voltage, activation overpotential, ohmic overpotential, concentration overpotential, and double-layer equivalent voltage.

[0008] The neural network control module is used to preprocess and feature process the fuel cell thermal management data collected by the sensor data collection module. The preprocessing includes data cleaning and outlier removal of the fuel cell thermal management data, and the feature processing includes feature selection, cooling, normalization and feature extraction of the fuel cell thermal management data. It is also used for training and updating the neural network control module itself.

[0009] The integrated thermal management system module is used to control the fuel cell stack system, radiator, heat exchanger, circulating water pump, cockpit, battery pack and heating device in the fuel cell vehicle based on the internal parameters obtained by feature processing and training updates of the neural network control module.

[0010] The sensor data collection module includes several temperature sensors and Hall sensors, which are respectively installed in the fuel cell vehicle's stack system, radiator, circulating water pump, heat exchanger, battery pack, coolant pipes, air conditioning pipes, cockpit, fuel cell stack system inlet and outlet, and heating device.

[0011] The specific process of training and updating the neural network control module itself is as follows:

[0012] Determine the number of nodes in the input layer, hidden layer, and output layer of the neural network control module, as well as the connection method between each network layer, and select the specific loss function and activation function according to the specific requirements of the integrated thermal management system module;

[0013] The fuel cell thermal management data is divided into training, validation, and test sets. The training and validation sets share 2 / 3 of the fuel cell thermal management data, while the test set uses 1 / 3 of the fuel cell thermal management data.

[0014] The neural network control module is trained using data from the training set. The specific process is as follows: the internal parameters of the neural network control module are gradually adjusted using the gradient descent optimization algorithm so that the prediction results of the module are close to the true values. The internal parameters include the membrane water content and internal humidity of the fuel cell system. During the training process, the learning rate, batch size and number of training iterations are set as hyperparameters.

[0015] The performance of the trained neural network control module was evaluated using a validation set, with the mean absolute error as the evaluation metric. This measured the prediction accuracy and fit of the module in the integrated thermal management system. Based on the validation results, the hyperparameters of the neural network control module were adjusted again to further optimize the neural network control module.

[0016] The neural network control module is evaluated using a test set. By predicting unknown data and comparing it with the true values, the module's performance and reliability in practical applications are assessed. Based on the comparison results, the network structure is further adjusted, the amount of training data is increased, and regularization steps are introduced to improve the module's performance and generalization ability.

[0017] The radiator includes a radiator body and a cooling fan. The radiator body contains continuous curved water channels, with coolant pipes connected to these channels. Coolant dissipates heat through the curved water channels and their walls. During this heat dissipation process, the cooling fan draws cool air into the curved water channel walls for rapid cooling. When the heat dissipated by the fuel cell stack system diminishes the radiator's continuous heat dissipation performance, a neural network control module controls the opening of a three-way valve in the integrated thermal management system module. This allows heat from the fuel cell stack system to be transferred to the heating device via coolant. The module also controls the opening of the heat exchanger valves, allowing some coolant to enter the heat exchanger, increasing heat dissipation. Heat from the heat exchanger is then transferred to the battery pack, reducing power loss from heating the battery pack under low-temperature conditions.

[0018] The integrated thermal management system module works as follows: it controls the fuel cell stack system, radiator, heat exchanger, circulating water pump, cockpit, battery pack, and heating device in the fuel cell vehicle based on the electrical signals of the internal parameters obtained through feature processing and training updates of the neural network control module. The internal parameters include the coolant temperature at the inlet and outlet of the fuel cell stack system. The received electrical signals are input to the circulating water pump and radiator to control the speed of the cooling fan in the radiator and drive the coolant in the cooling pipes of the circulating water pump to circulate in the cooling pipes, thereby reducing heat accumulation in the fuel cell stack system and reducing the impact of temperature on the humidity and temperature of the proton exchange membrane inside the fuel cell stack system.

[0019] When the operating temperature of the fuel cell stack system is below the first temperature threshold, the circulating water pump is controlled to reduce its speed, so that the coolant circulates only in the circulating water pump and the fuel cell stack system, thereby reducing the energy loss of the entire vehicle. When the operating temperature of the fuel cell stack system exceeds the second temperature threshold, the opening of the three-way valve is controlled to increase, so that the coolant circulates in the circulating water pump, the water tank in the integrated thermal management system module, the three-way valve, and the radiator, thereby reducing the heat accumulation in the fuel cell stack system and controlling the operating temperature of the fuel cell stack system between the first temperature threshold and the second temperature threshold. The range between the first temperature threshold and the second temperature threshold is the optimal operating temperature range of the fuel cell stack system.

[0020] The first temperature threshold is 60℃, and the second temperature threshold is 80℃.

[0021] The integrated thermal management system module controls the cockpit, battery pack, and heating device in the following ways: To reduce the risk of damage to the radiator caused by prolonged operation, the module controls the opening of the three-way valve and the heat exchanger valve based on the electrical signals of the internal parameters obtained through feature processing and training updates of the neural network control module. This allows some heat to be transferred to the battery pack through the heat exchanger, reducing the amount of heat required to maintain the battery pack's operating temperature. Additionally, some heat is transferred to the heating device through the coolant, reducing the energy consumption of the heating device, and some heat is transferred to the cockpit.

[0022] The integrated thermal management system module also controls the cockpit and heating device in the following ways: under low temperature conditions, the fuel cell stack system heats up slowly. Based on the electrical signals of the internal parameters obtained by the neural network control module through feature processing and training updates, a low temperature warning signal is issued to the cockpit. This allows the system to heat the fuel cell stack system through the warm air delivery pipes of the air conditioning system in the fuel cell vehicle while heating the cockpit, thus achieving rapid temperature rise, reducing the initial cold start time of the fuel cell stack, and reducing the power consumption of the entire vehicle.

[0023] A neural network-based integrated thermal management method for fuel cell vehicles is also provided, comprising the following steps:

[0024] Collect fuel cell thermal management data, including the temperature of the fuel cell stack system, the temperature of the cooling system, the temperature of the cabin, the voltage and current values ​​of the battery pack, and the ambient temperature of the fuel cell vehicle, in order to obtain the thermal state of the stack, the performance of the cooling system, the working status of the battery pack, and the current output.

[0025] The various operating processes and parameter contents inside the fuel cell stack system are converted into specific values. The structure of the fuel cell stack system and the converted specific values ​​are displayed. The output characteristics, efficiency and temperature distribution of the fuel cell stack system under different operating conditions are simulated. The operating process includes the generation process of reversible voltage, activation overpotential, ohmic overpotential, concentration overpotential and double-layer equivalent voltage.

[0026] The fuel cell thermal management data collected by the sensor data collection module is preprocessed and feature-processed. The preprocessing includes data cleaning and outlier removal of the fuel cell thermal management data, and the feature processing includes feature selection, cooling, normalization and feature extraction of the fuel cell thermal management data; the neural network control module itself is trained and updated.

[0027] The internal parameters obtained through feature processing and training updates of the neural network control module are used to control the fuel cell stack system, radiator, heat exchanger, circulating water pump, cockpit, battery pack, and heating device in the fuel cell vehicle.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] This invention achieves temperature control of fuel cells through a more precise integrated thermal management system module, thereby achieving higher energy transfer efficiency and effective power output of the fuel cell stack. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the structure of an embodiment of the present invention;

[0031] Figure 2 This is a schematic diagram showing the connection between the neural network control module and the integrated thermal management system module in an embodiment of the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0033] This invention provides an integrated thermal management system for fuel cell vehicles based on neural networks, such as... Figure 1 As shown, the thermal management system of the present invention includes a sensor data collection module, an electric stack module, a neural network control module, and an integrated thermal management system module.

[0034] The sensor data collection module requires various sensors that are installed on the fuel cell vehicle stack, radiator, circulating water pump, heat exchanger, battery pack, coolant pipes, air conditioning pipes, cockpit, fuel cell stack inlet / outlet, and hydrogen heating device side. These sensors are used to sense and acquire in real time the thermal state of the fuel cell system, the performance of the cooling system, the cockpit temperature, the working status and current output of the battery pack, and the external environmental conditions of the fuel cell vehicle.

[0035] The fuel cell stack module transforms the various operating processes and parameter contents inside the fuel cell stack into specific simulation modules, intuitively displaying the structure and values ​​of the entire fuel cell stack system, and simulating the output characteristics, efficiency, temperature distribution, etc. of the fuel cell stack under different operating conditions.

[0036] The neural network control module is used to train the sensor data collection module and verify the trained module to achieve high-efficiency operation of the integrated thermal management system.

[0037] The neural network control module is described in detail. First, it needs to collect data related to fuel cell thermal management, including parameters such as temperature, pressure, and flow rate of the fuel cell system, as well as external environmental parameters such as temperature and humidity. This is to ensure data quality and requires preprocessing, such as data cleaning and outlier removal. These parameters originate from the sensor data collection module. Next, feature selection is required. Based on the stack temperature requirements and the characteristics of the sensor-collected data, feature engineering is performed on the raw data to extract useful features and remove abrupt temperature and humidity fluctuations. This step employs feature selection, dimensionality reduction, and normalization methods to improve the module's performance and generalization ability. Then, the neural network module is constructed, including determining the number of nodes in the input, hidden, and output layers, as well as the connection methods between network layers. Specific loss and activation functions are selected based on the specific requirements of the thermal management system. Considering the actual internal temperature of the fuel cell stack, this invention selects a convolutional neural network for better regression and classification tasks. Finally, data partitioning and module evaluation testing are carried out. This part requires dividing the data collected in the sensor data collection module into training set, validation set and test set. The training set and validation set use 2 / 3 of the data under actual working conditions, and the test set uses 1 / 3 of the data under actual working conditions.

[0038] The neural network module is trained using data from the training set. The main process involves gradually adjusting the internal parameters of the neural network module using optimization algorithms such as gradient descent. The key parameters are the water content of the fuel cell membrane and the internal humidity of the fuel cell, with the aim of making the module's predictions approximate the true values. Hyperparameters such as learning rate, batch size, and number of training iterations are set during the training process.

[0039] The performance of the trained module is evaluated using a validation set, and the mean absolute error is used as the metric to measure the module’s prediction accuracy and fit on the thermal management system. Based on the validation results, the hyperparameters of the neural network module are readjusted to further optimize the module.

[0040] The final neural network control module is evaluated using a test set. By predicting unknown data and comparing it with the actual values ​​under engineering conditions, the module's performance and reliability in practical applications are assessed. Based on the comparison results, the network structure is further adjusted, the amount of training data is increased, and regularization steps are introduced to improve the module's performance and generalization ability.

[0041] like Figure 2 As shown, the integrated thermal management system module is used to execute the internal parameter values ​​and control strategies transmitted by the neural network control module, thereby controlling the fuel cell stack and components such as radiators, heat exchangers, and circulating water pumps, and controlling the cockpit, battery packs, and hydrogen heaters.

[0042] The integrated thermal management system module is described in detail. This module receives control strategies and pre-adjusted internal parameter electrical signals from other modules. These parameters primarily include the coolant temperature at the fuel cell inlet and outlet. The received electrical signals are input to the circulating water pump module and the radiator module to control the speed of their respective internal fans. The circulating water pump drives the coolant in the cooling pipes to circulate within the cooling loop, reducing heat buildup inside the fuel cell and minimizing the impact of temperature on the humidity and temperature of the proton exchange membrane inside the fuel cell. The radiator controls the heat dissipation of the cooling fans based on the magnitude of the received electrical signals.

[0043] To further explain, the radiator consists of a radiator body and a cooling fan. The radiator body is composed of curved water channels, through which coolant conducts heat away. During this heat transfer, the cooling fan draws cool air from the air into the water channel walls, achieving rapid cooling. When the fuel cell stack operates at high power for extended periods, the continuous heat dissipation can degrade the radiator's heat dissipation performance. Based on electrical signals from the neural network control module, the opening of the three-way valve is controlled, allowing heat from inside the fuel cell stack to be transferred to the hydrogen heating device via coolant. The opening of the heat exchanger valve is also controlled, allowing some coolant to enter the heat exchanger piping, increasing heat dissipation. Heat from the heat exchanger is then transferred to the battery pack, reducing power consumption for heating the battery pack under low-temperature conditions.

[0044] To further explain, the optimal operating temperature for mainstream fuel cell stacks on the market is currently 60-80℃. When the fuel cell stack operating temperature is below 60℃, only the circulating water pump is controlled to run at a low speed, allowing the coolant to circulate slightly between the pump and the stack, reducing overall vehicle energy loss. When the fuel cell stack temperature exceeds 80℃, the opening of the three-way valve is increased, allowing the coolant to enter the main circulation loop through the pump, water tank, three-way valve, and radiator. This aims to reduce heat buildup in the fuel cell stack and keep its temperature below 80℃.

[0045] To further explain, when the vehicle is in a high-power operating state for a long time, in order to reduce the risk of damage to the radiator caused by prolonged operation, the opening of the three-way valve and the heat exchanger valve is controlled based on the feedback signal from the neural network module. This allows some heat to be transferred to the battery pack through the heat exchanger, reducing the heat required to maintain the operating temperature of the battery pack. Additionally, some heat is transferred to the hydrogen heating device through the coolant, reducing the heat required for hydrogen heating, and some heat is transferred to the cabin, thus maintaining the cabin temperature.

[0046] To further explain, under low-temperature conditions, the fuel cell stack heats up slowly. Based on the feedback signal from the neural network module, a low-temperature warning signal is issued to the dashboard in the driver's cab. Then, based on the driver's cab signal, the air conditioning system is controlled to heat the driver's cab while simultaneously heating the fuel cell stack through the warm air delivery pipes, which can achieve rapid temperature rise, reduce the initial cold start time of the fuel cell stack, and reduce the power consumption and hydrogen consumption of the entire vehicle.

[0047] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An integrated thermal management system for fuel cell vehicles based on neural networks, characterized in that, include: The sensor data collection module is used to collect fuel cell thermal management data, including the temperature of the fuel cell stack system, the temperature of the cooling system, the temperature of the cabin, the voltage and current values ​​of the battery pack, and the ambient temperature of the fuel cell vehicle, in order to obtain the thermal state of the stack, the performance of the cooling system, the operating status of the battery pack, and the current output. The fuel cell stack module is used to convert various operating processes and parameter contents within the fuel cell stack system into specific values, display the structure of the fuel cell stack system and the converted specific values, and simulate the output characteristics, efficiency, and temperature distribution of the fuel cell stack system under different operating conditions. The operating process includes the generation process of reversible voltage, activation overpotential, ohmic overpotential, concentration overpotential, and double-layer equivalent voltage. The neural network control module is used to preprocess and feature process the fuel cell thermal management data collected by the sensor data collection module. The preprocessing includes data cleaning and outlier removal of the fuel cell thermal management data, and the feature processing includes feature selection, cooling, normalization and feature extraction of the fuel cell thermal management data. It is also used for training and updating the neural network control module itself. An integrated thermal management system module is used to control the fuel cell stack system, radiator, heat exchanger, circulating water pump, cockpit, battery pack and heating device in fuel cell vehicles based on the internal parameters obtained by feature processing and training updates of the neural network control module. The integrated thermal management system module controls the cockpit, battery pack, and heating device as follows: To reduce the risk of damage to the radiator caused by prolonged operation, it controls the valve opening of the three-way valve and the heat exchanger valve based on the electrical signals of the internal parameters obtained by the neural network control module's feature processing and training update. This allows some heat to be transferred to the battery pack through the heat exchanger, reducing the amount of heat required to maintain the battery pack's operating temperature. Additionally, some heat is transferred to the heating device through the coolant, reducing the heating device's energy consumption, and some heat is transferred to the cockpit. The integrated thermal management system module also controls the cockpit and heating device in the following ways: under low temperature conditions, the fuel cell stack system heats up slowly. Based on the electrical signals of the internal parameters obtained by the neural network control module through feature processing and training updates, a low temperature warning signal is issued to the cockpit. This allows the system to heat the fuel cell stack system through the warm air delivery pipes of the air conditioning system in the fuel cell vehicle while heating the cockpit, thus achieving rapid temperature rise, reducing the initial cold start time of the fuel cell stack, and reducing the power consumption of the entire vehicle.

2. The integrated thermal management system for fuel cell vehicles based on neural networks according to claim 1, characterized in that, The sensor data collection module includes several temperature sensors and Hall sensors, which are respectively installed in the fuel cell vehicle's stack system, radiator, circulating water pump, heat exchanger, battery pack, coolant pipes, air conditioning pipes, cockpit, fuel cell stack system inlet and outlet, and heating device.

3. The integrated thermal management system for fuel cell vehicles based on neural networks according to claim 1, characterized in that, The specific process of training and updating the neural network control module itself is as follows: Determine the number of nodes in the input layer, hidden layer, and output layer of the neural network control module, as well as the connection method between each network layer, and select the specific loss function and activation function according to the specific requirements of the integrated thermal management system module; The fuel cell thermal management data is divided into training, validation, and test sets. The training and validation sets share 2 / 3 of the fuel cell thermal management data, while the test set uses 1 / 3 of the fuel cell thermal management data. The neural network control module is trained using data from the training set. The specific process is as follows: the internal parameters of the neural network control module are gradually adjusted using the gradient descent optimization algorithm so that the prediction results of the module are close to the true values. The internal parameters include the membrane water content and internal humidity of the fuel cell system. During the training process, the learning rate, batch size and number of training iterations are set as hyperparameters. The performance of the trained neural network control module was evaluated using a validation set, with the mean absolute error as the evaluation metric. This measured the prediction accuracy and fit of the module in the integrated thermal management system. Based on the validation results, the hyperparameters of the neural network control module were adjusted again to further optimize the neural network control module. The neural network control module is evaluated using a test set. By predicting unknown data and comparing it with the true values, the module's performance and reliability in practical applications are assessed. Based on the comparison results, the network structure is further adjusted, the amount of training data is increased, and regularization steps are introduced to improve the module's performance and generalization ability.

4. The integrated thermal management system for fuel cell vehicles based on neural networks according to claim 2, characterized in that, The radiator includes a radiator body and a cooling fan. The radiator body contains continuous curved water channels, with coolant pipes connected to these channels. Coolant dissipates heat through the curved water channels and their walls. During this heat dissipation process, the cooling fan draws cool air into the curved water channel walls for rapid cooling. When the heat dissipated by the fuel cell stack system diminishes the radiator's continuous heat dissipation performance, a neural network control module controls the opening of a three-way valve in the integrated thermal management system module. This allows heat from the fuel cell stack system to be transferred to the heating device via coolant. The module also controls the opening of the heat exchanger valves, allowing some coolant to enter the heat exchanger, increasing heat dissipation. Heat from the heat exchanger is then transferred to the battery pack, reducing power loss from heating the battery pack under low-temperature conditions.

5. The integrated thermal management system for fuel cell vehicles based on neural networks according to claim 4, characterized in that, The integrated thermal management system module works as follows: it controls the fuel cell stack system, radiator, heat exchanger, circulating water pump, cockpit, battery pack, and heating device in the fuel cell vehicle based on the electrical signals of the internal parameters obtained through feature processing and training updates of the neural network control module. The internal parameters include the coolant temperature at the inlet and outlet of the fuel cell stack system. The received electrical signals are input to the circulating water pump and radiator to control the speed of the cooling fan in the radiator and drive the coolant in the cooling pipes of the circulating water pump to circulate in the cooling pipes, thereby reducing heat accumulation in the fuel cell stack system and reducing the impact of temperature on the humidity and temperature of the proton exchange membrane inside the fuel cell stack system.

6. The integrated thermal management system for fuel cell vehicles based on neural networks according to claim 4, characterized in that, When the operating temperature of the fuel cell stack system is below the first temperature threshold, the circulating water pump is controlled to reduce its speed, so that the coolant circulates only in the circulating water pump and the fuel cell stack system, thereby reducing the energy loss of the entire vehicle. When the operating temperature of the fuel cell stack system exceeds the second temperature threshold, the opening of the three-way valve is controlled to increase, so that the coolant circulates in the circulating water pump, the water tank in the integrated thermal management system module, the three-way valve, and the radiator, thereby reducing the heat accumulation in the fuel cell stack system and controlling the operating temperature of the fuel cell stack system between the first temperature threshold and the second temperature threshold. The range between the first temperature threshold and the second temperature threshold is the optimal operating temperature range of the fuel cell stack system.

7. The integrated thermal management system for fuel cell vehicles based on neural networks according to claim 6, characterized in that, The first temperature threshold is 60℃, and the second temperature threshold is 80℃.

8. A method for using a neural network-based integrated thermal management system for fuel cell vehicles as described in any one of claims 1-7, characterized in that, Includes the following steps: Collect fuel cell thermal management data, including the temperature of the fuel cell stack system, the temperature of the cooling system, the temperature of the cabin, the voltage and current values ​​of the battery pack, and the ambient temperature of the fuel cell vehicle, in order to obtain the thermal state of the stack, the performance of the cooling system, the operating status of the battery pack, and the current output. The various operating processes and parameter contents inside the fuel cell stack system are converted into specific values. The structure of the fuel cell stack system and the converted specific values ​​are displayed. The output characteristics, efficiency and temperature distribution of the fuel cell stack system under different operating conditions are simulated. The operating process includes the generation process of reversible voltage, activation overpotential, ohmic overpotential, concentration overpotential and double-layer equivalent voltage. The fuel cell thermal management data collected by the sensor data collection module is preprocessed and feature-processed. The preprocessing includes data cleaning and outlier removal of the fuel cell thermal management data, and the feature processing includes feature selection, cooling, normalization and feature extraction of the fuel cell thermal management data; the neural network control module itself is trained and updated. The internal parameters obtained through feature processing and training updates of the neural network control module are used to control the fuel cell stack system, radiator, heat exchanger, circulating water pump, cockpit, battery pack, and heating device in the fuel cell vehicle.

Images