A thermal management system and temperature control method for a proton exchange membrane fuel cell

Abstract

This invention discloses a thermal management system and temperature control method for a proton exchange membrane fuel cell. The thermal management system includes a fuel cell stack, a particulate filter, a three-way valve, a heater, a main radiator, a secondary radiator, a fan, a water pump, an expansion tank, and a control module. The cooling water circulation loop can be divided into a heating branch and a cooling branch to meet the system requirements at different operating stages. In this invention, the proton exchange membrane fuel cell thermal management system reuses the cooling air blown in by the fan. Because there is still a 30K temperature difference between the cooling air passing through the main radiator and the cooling water in the secondary radiator, it can be used to cool the secondary radiator again. This reduces the required cooling airflow and allows the fan to move out of its inefficient zone, significantly reducing the parasitic power consumed by the fan. The control method of this invention includes: the control module uses two independent PID controllers to adjust the speed of the water pump and the fan respectively to control the outlet and inlet temperatures of the cooling water. Furthermore, the PID controller on the water pump side is optimized through reinforcement learning, improving the adaptive capability of the thermal management system and enhancing the control effect.

Description

Technical Field

[0001] This invention belongs to the field of proton exchange membrane fuel cell technology, specifically relating to a thermal management system and temperature control method for a proton exchange membrane fuel cell. Background Technology

[0002] Proton exchange membrane fuel cells (PEMFCs) can directly convert chemical energy into electrical energy, breaking free from the Carnot cycle limitations of traditional heat engines and exhibiting higher energy conversion efficiency. At the same time, PEMFCs are characterized by high power density, low operating temperature, and environmental friendliness, and are considered to be highly promising future power generation devices.

[0003] Temperature is a major factor affecting the performance of PEMFCs. Increased temperature can accelerate chemical reaction rates, but excessively high temperatures can cause proton exchange membrane dehydration, reduce proton conductivity, and even cause irreversible damage. Conversely, excessively low temperatures can accelerate water vapor condensation, leading to flooding and increasing gas mass transfer resistance. Therefore, it is necessary to control the appropriate operating temperature of the PEMFC to improve its performance and maintain efficient and stable operation. Furthermore, the inlet and outlet temperature difference of the PEMFC must be controlled within a given range; otherwise, excessive thermal stress will accelerate the aging of the proton exchange membrane and shorten the lifespan of the PEMFC. Typically, battery stack temperature measurement is indirect; therefore, in practical engineering, the outlet cooling water temperature is often used to represent the battery stack temperature. In summary, PEMFCs require an external thermal management system to control the inlet and outlet cooling water temperatures to maintain a suitable battery stack operating temperature and inlet / outlet temperature difference, ensuring efficient and stable system operation. In addition, within the thermal management system, a small temperature difference between the cooling water in the radiator and the ambient temperature can lead to excessively high cooling fan power, becoming a significant factor limiting the system's net output power. For example, commercial heavy-duty vehicles need to consume 8% or even more of the total system power to meet heat dissipation requirements, resulting in a huge parasitic power loss.

[0004] Currently, classical PID control is the most widely used method for temperature control in PEMFC (Pre-Earth Flow Control System), and its control effect mainly depends on three fixed control parameters, Kp, Ki, and Kd. However, PEMFC is a complex system with nonlinearity, strong coupling, and large time delay, and classical PID cannot achieve optimal control performance. In this context, some researchers have introduced fuzzy control to optimize classical PID, enabling adaptive changes in Kp, Ki, and Kd during the control process. The selection of the membership function and the formulation of fuzzy rules in fuzzy control have a significant impact on the control effect, but the relevant parameters can only be obtained based on expert and engineering experience. Therefore, fuzzy control often fails to achieve satisfactory control results. Summary of the Invention

[0005] To address the problems in the prior art, this invention provides a thermal management system and temperature control method for proton exchange membrane fuel cells, solving the problem of the huge parasitic power of the cooling fan affecting the net power and efficiency of the entire system. At the same time, this invention combines reinforcement learning and PID control to adaptively change Kp, Ki, and Kd, and compares the control effect with that of general fuzzy PID. The results show that reinforcement learning PID can effectively improve the temperature control effect and enhance the system's adaptability.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0007] A thermal management system for a proton exchange membrane fuel cell, referred to as a PEMFC system, includes:

[0008] Fuel cell stacks are used to convert the chemical energy in hydrogen into electrical energy.

[0009] The cooling water circulation loop includes a water pump, a heater, an expansion tank, a main radiator, and an auxiliary radiator; the output end of the water pump is connected to the cooling water inlet of the fuel cell stack; the cooling water outlet of the fuel cell stack is connected to the inlet of the heater and the inlet of the auxiliary radiator respectively through the two output ends of a three-way valve; the outlets of the heater and the main radiator are connected to the input end of the water pump; the cooling water circulation loop is divided into a heating branch and a cooling branch.

[0010] The heating branch is used to preheat the cooling water during the startup phase of the PEMFC system.

[0011] The heat dissipation branch is used to transfer the heat generated by the fuel cell stack during the normal operation of the PEMFC system, control the inlet and outlet temperatures of the cooling water, and maintain the efficient and stable operation of the system.

[0012] The control module is used to control the three-way valve to connect with the heating branch and disconnect the heat dissipation branch when the PEMFC system is in the startup phase, and to turn on the heater to preheat the cooling water to the given temperature; when the PEMFC system is in the normal operation phase, it controls the three-way valve to connect with the heat dissipation branch and disconnect the heating branch, turns off the heater, and controls the cooling water outlet temperature and inlet temperature respectively through two independent PID controllers.

[0013] A further improvement of the present invention is that the heating branch includes a fuel cell stack particulate filter, a three-way valve, a heater, an expansion tank and a water pump, wherein the expansion tank is used to provide cooling water and play a role in stabilizing pressure and temperature.

[0014] After the cooling water flows through the particulate filter, it is guided to the heater through a three-way valve for heating, and then flows to the fuel cell stack through a water pump. After the cooling water flows through the fuel cell stack, it flows back through the particulate filter for a new cycle.

[0015] A further improvement of the present invention is that the heat dissipation branch includes a PEMFC, a particulate filter, a three-way valve, a main radiator, a secondary radiator, an expansion tank, and a water pump, wherein the expansion tank is used to provide cooling water and play a role in stabilizing pressure and temperature.

[0016] The main radiator and the auxiliary radiator are connected by a closed channel; after the cooling water flows through the particulate filter, it is guided to the auxiliary radiator for heat dissipation through a three-way valve, then flows through the main radiator for further heat dissipation, and then flows to the fuel cell stack through a water pump; after the cooling water flows through the fuel cell stack, it flows back through the particulate filter for a new cycle.

[0017] A further improvement of the present invention is that, during operation, it includes:

[0018] During the PEMFC system startup phase, due to the low temperature of the cooling water, in order to prevent the cooling water from carrying away the heat of the battery stack and causing excessive hydrogen consumption, the cooling water is preheated to the set temperature through the heating branch.

[0019] During the normal operation of the PEMFC system, the battery stack generates a large amount of heat during operation and controls the inlet and outlet temperatures of the cooling water. At this time, the cooling water flows through the heat dissipation branch to dissipate heat from the system.

[0020] A temperature control method for a proton exchange membrane fuel cell, the method being based on a thermal management system for a proton exchange membrane fuel cell, includes the following steps:

[0021] S1: Establish a semi-empirical voltage model for PEMFC to obtain the output power of the battery stack;

[0022] S2: Based on the battery stack output power obtained from S1, and combined with the physical characteristics of the water pump, radiator and expansion tank, a PEMFC thermal management system model is constructed according to the energy conservation equation, and built on the Matlab / Simulink platform to obtain the temperature change of the PEMFC system.

[0023] S3: Based on the fan characteristic curve, construct a fan power model to study the fan parasitic power problem, which is used to calculate the control energy consumption of the controller established in S4 during the control process;

[0024] S4: Based on the PEMFC thermal management system model established in S2, two independent PID controllers are established to adjust the speed of the fan and water pump respectively to control the inlet and outlet temperatures of the cooling water.

[0025] S5: Optimize the pump-side PID controller established in S4 using reinforcement learning to achieve K... p K i and K dAdaptive changes during the control process.

[0026] A further improvement of the present invention is that, in step S1, the output power of the fuel cell stack is calculated by the following equation:

[0027] V cell =e nerst -η act -η ohm -η conc

[0028] V stack =N·V cell

[0029] P st =I·V stack

[0030] In the formula, n nerst Represents the reversible voltage, η act η represents the voltage loss caused by the activation of the electrochemical reaction. ohmic η represents the voltage loss caused by ion and electron conduction. conc V represents the voltage loss caused by the diffusion of reactant concentration. cell V represents the voltage of a single battery cell. stack P represents the battery stack voltage. st This represents the output power of the battery stack, where N is the number of battery cells.

[0031] A further improvement of the present invention is that, in step S2, the energy balance equation for constructing the PEMFC thermal management system is as follows:

[0032]

[0033] In the formula, C st M represents the specific heat capacity of the PEMFC stack. st T represents the mass of the fuel cell stack. st Q represents the temperature of the fuel cell stack. tot P represents the total chemical energy of the reactants. st Q represents the output power of the fuel cell stack. gas Q represents the net heat carried out of the system by the inlet and outlet gases. cl Q represents the heat carried out of the system by the cooling water. atm This represents the heat emitted by the system into the environment.

[0034] A further improvement of the present invention is that, in step S3, the fan power in the fan power model is calculated by the following equation:

[0035]

[0036] In the formula, N represents the actual fan speed, N0 represents the maximum fan speed, and q air p represents airflow rate s η represents the static pressure of the fan. fan p represents the efficiency of the fan. s and η fan All are airflow rates q air The function is used to obtain the p value in the fan characteristic curve through curve fitting. s and η fan The equation of the curve.

[0037] A further improvement of this invention is that, in step S4, the outlet cooling water temperature is controlled by adjusting the water pump speed, and the inlet cooling water temperature is controlled by adjusting the fan speed. The PID controller satisfies the following equation in the time domain:

[0038]

[0039] Among them, K p K is the proportionality coefficient. i K is the integral coefficient. d The differential coefficient is denoted by , and the error function e(t) represents the difference between the real-time temperature of the PEMFC and the target temperature.

[0040] A further improvement of this invention is that, in step S5, the reinforcement learning method employed includes two parts: the environment and the agent. When the agent interacts with the environment, the following events occur at each time t:

[0041] (1) The environmental state s(t) perceived by the agent at time t;

[0042] (2) Based on the current state s(t) and reinforcement learning information, the system selects and executes an action a(t);

[0043] (3) Action a(t) acts on the current environment, the environment changes, and enters a new state, i.e. s(t)→s(t+1);

[0044] (4) System feedback evaluation function;

[0045] (5) The agent accepts the evaluation function and changes the reinforcement information. The system returns to the previous step, i.e., s(t+1)→s(t);

[0046] (6) Return to step (1) and repeat the above steps until a satisfactory system state is obtained, then end the loop;

[0047] A PID controller based on reinforcement learning includes a performance evaluation unit, a parameter correction unit, and a decision unit.

[0048] The performance evaluation unit evaluates the current control effect and generates corresponding adjustment amounts based on the evaluation function;

[0049] The parameter correction unit corrects Kp, Ki, and Kd based on the adjustment amount of the performance evaluation unit;

[0050] The decision-making unit finds a set of optimal values ​​from the adjustment amount, that is, obtains the adjustment amount with the optimal evaluation function value.

[0051] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0052] (1) This invention provides a thermal management system for a proton exchange membrane fuel cell. By reusing the cooling air blown in by the fan, the cooling air flow can be reduced, and the fan can be moved out of the inefficient operating zone, effectively reducing the fan's parasitic power and improving the efficiency of the PEMFC system. At the same time, the thermal management system of the proton exchange membrane fuel cell can improve the control effect of the inlet cooling water temperature to a certain extent;

[0053] (2) The present invention provides a temperature control method for proton exchange membrane fuel cells. The method is based on a reinforcement learning PID controller. The reinforcement learning method is used to optimize the PID controller on the pump side, which improves the system's adaptability and improves the control effect of the outlet cooling water temperature. Attached Figure Description

[0054] Figure 1 This is a structural diagram of the PEMFC thermal management system;

[0055] Figure 2 This is a structural diagram of the temperature control method of the PEMFC thermal management system;

[0056] Figure 3 It is the fan characteristic curve;

[0057] Figure 4 This is a diagram of a PID controller based on reinforcement learning.

[0058] Figure 5 This is a schematic diagram of the current disturbance applied to the PEMFC thermal management system;

[0059] Figure 6 This is a comparison chart of the cooling water outlet temperature control effect during the operation of the PEMFC system;

[0060] Figure 7 This is a schematic diagram showing the changes in control parameters during the control process of a fuzzy PID controller.

[0061] Figure 8 This is a schematic diagram illustrating the changes in control parameters during the control process of a reinforcement learning PID controller.

[0062] Figure 9 This is a comparison chart of cooling airflow during the operation of the PEMFC system;

[0063] Figure 10 This is a comparison chart showing the cooling water inlet temperature control effect during the operation of the PEMFC system.

[0064] Explanation of icon numbers:

[0065] 1 is the fuel cell stack, 2 is the particulate filter, 3 is the three-way valve, 4 is the auxiliary radiator, 5 is the main radiator, 6 is the fan, 7 is the heater, 8 is the water pump, 9 is the expansion tank, and 10 is the control module. Detailed Implementation

[0066] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0067] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0068] In the description of this invention, the use of "first" and "second" is for the purpose of distinguishing technical features only, and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the order of the technical features indicated.

[0069] In the description of this invention, unless otherwise explicitly defined, terms such as "set up," "install," and "connect" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0070] This specific embodiment discloses a thermal management system for a proton exchange membrane fuel cell, which mainly includes the following:

[0071] The fuel cell stack 1, cooling water circulation loop (including heating and cooling branches), and control module 10 are included. The cooling water circulation loop includes a particulate filter 2, an auxiliary radiator 4, a main radiator 5, a heater 7, a water pump 8, and an expansion tank 9. The output of the water pump is connected to the cooling water inlet of the fuel cell stack. The cooling water outlet of the fuel cell stack is connected to the inlet of the heater and the inlet of the auxiliary radiator respectively through the two outputs of a three-way valve. The outlets of the heater and the main radiator are connected to the input of the water pump. The heating branches in the cooling water circulation loop are connected at both ends to one output of the three-way valve and the loop after the main radiator. Each branch includes a heater 7. It serves to preheat the cooling water during the initial stage of system operation; the two ends of the heat dissipation branch in the cooling water circulation loop are respectively connected between the other output end of the three-way valve and the circuit after the heater. The main radiator 5 and the auxiliary radiator 4 in the branch are connected by a closed channel. The cooling air blown in by the fan 6 around the main radiator flows from the main radiator to the auxiliary radiator through the closed channel for secondary cooling; the expansion tank 9 mainly functions to stabilize pressure and temperature and provide cooling water; the control module 10 controls the three-way valve 3, the main radiator 5 and the water pump 8, and is used to control the flow direction of the three-way valve and the inlet and outlet temperatures of the cooling water in different working stages of the PEMFC system.

[0072] The heating branch pipe is filled with cooling water. When the heating branch is turned on, the cooling water flows out from the fuel cell stack 1, passes through the particulate filter 2, enters the heater 7 through the three-way valve 3 for heating, and then flows back to the stack through the water pump 8. When the system starts running for a period of time, the cooling water temperature is low, so the heating branch is turned on to preheat the cooling water, allowing the PEMFC system to enter the normal operation stage more quickly.

[0073] The cooling branch pipes are filled with cooling water. The cooling water flows out of the fuel cell stack 1, passes through the particulate filter 2, then through the three-way valve 3, first entering the auxiliary radiator 4 for initial cooling, then entering the main radiator 5 for further cooling, and finally flowing back to the stack via the water pump 8. In the system model disclosed in this patent, secondary cooling is achieved using the nearly 30K temperature difference between the cooling air and cooling water passing through the main radiator. The main implementation method involves setting up an auxiliary radiator and establishing a closed channel between the main and auxiliary radiators. Cooling air passing through the main radiator is then diverted to the auxiliary radiator through this closed channel. This secondary cooling method allows for the removal of more heat with the same cooling airflow, thereby reducing the cooling airflow and keeping the fan away from its inefficient operating zone, effectively reducing fan parasitic power. When the system is operating normally, to maintain a suitable PEMFC operating temperature, the cooling water removes the heat generated by the fuel cell stack and flows through the cooling branch for cooling, thus ensuring efficient and stable system operation.

[0074] The particulate filter is connected between the fuel cell stack cooling water outlet and the three-way valve. Its function is to filter impurities and contaminants in the cooling water and prevent blockage of the cooling water circuit.

[0075] The outlet of the expansion tank is connected to the input of the water pump, and the inlet of the expansion tank is connected to the outlet of the heater and the main radiator. The expansion tank mainly functions to stabilize pressure and temperature. Utilizing the space above it, the compressibility of air helps maintain system stability; it also discharges gas from the system, separating gas and liquid to ensure stable internal pressure. Furthermore, the expansion tank provides cooling water and acts as a heat capacity, reducing temperature fluctuations in the outlet cooling water and thus stabilizing the temperature.

[0076] The main function of the control module 10 is to better control the cooling water temperature. In the model disclosed in this patent, the target outlet water temperature is 343K, the inlet water temperature is 333K, and the heating branch temperature threshold is set to 333K. When the PEMFC is in the startup phase, the cooling water temperature is lower than the set heating branch temperature threshold. The control module controls the three-way valve to open the heating branch, allowing cooling water to flow through the heater and increase the cooling water temperature. When the PEMFC system is running normally, the cooling water temperature reaches (greater than or equal to) the set heating branch temperature threshold. The control module controls the three-way valve to close the heating branch, allowing cooling water to flow through the main and auxiliary radiators, which can promptly transfer the heat generated by the PEMFC system and achieve the temperature control target.

[0077] This specific embodiment discloses a PEMFC temperature control method based on reinforcement learning PID, including the following steps:

[0078] S1: Establish a semi-empirical PEMFC model to obtain the output power of the battery stack;

[0079] S2: Combining the physical characteristics of the water pump, radiator, and expansion tank, a PEMFC thermal management system model is constructed based on the energy conservation equation, and built on the Matlab / Simulink platform to obtain the temperature change of the PEMFC system.

[0080] S3: Based on the fan characteristic curve, construct a fan power model and study the fan parasitic power problem;

[0081] S4: Establish two PID controllers to control the inlet and outlet temperatures of the cooling water respectively;

[0082] S5: Reinforcement learning is used to optimize the PID control on the pump side, enabling adaptive changes in Kp, Ki, and Kd during the control process, thereby improving the control effect.

[0083] In step S1, the output voltage of the PEMFC is calculated by the following equation:

[0084] V cell =E nerst -η act -η ohmic -η conc ,

[0085] In the formula, E nerst Represents the reversible voltage, η act η represents the voltage loss caused by the activation of the electrochemical reaction. ohmic For ohmic loss, representing the voltage loss caused by ion and electron conduction, η conc Concentration loss represents the voltage loss caused by the concentration diffusion of reactants.

[0086] The reversible voltage is calculated according to the Nernst equation:

[0087]

[0088] In the formula, ΔS is the reaction entropy, n is the number of electrons transferred in the electrode reaction, F is the charge of a single electron, and T is the actual temperature. ref For reference temperature, p i This represents the partial pressure of component i.

[0089] Activation loss is calculated using the following empirical equation:

[0090]

[0091] In the formula, ξ1~ξ4 are semi-empirical equation parameters. This indicates the oxygen concentration in the battery stack.

[0092] The current is calculated by the following equation:

[0093] I = i + i n ,

[0094] Ohmic loss is calculated using Ohm's law:

[0095] η ohmic =IR int =I(R) m +R c ),

[0096] In the formula, R c The resistance for ion conduction; R m The resistance for electron conduction is calculated by the following equation:

[0097]

[0098] In the formula, ρ m The resistivity of the membrane is calculated using the following empirical equation:

[0099]

[0100] In the formula, λ represents the water content of the exchange membrane.

[0101] Concentration loss is calculated by the following equation:

[0102]

[0103] In the formula, J is actually the current density, J max Where b is the maximum current density and b is the concentration loss coefficient;

[0104] In the semi-empirical equation, the activation loss parameters ξ1~ξ4 and the ion conduction resistance R are... c Concentration loss coefficient b, membrane water content λ, maximum current density J max The current density J was obtained from experimental data.

[0105] In step S2, as Figure 1 As shown, the fuel cell system consists of a PEMFC stack, an expansion tank, a water pump, a radiator, and a fan. The heat generated by the stack is first transferred to the cooling water, and then the cooling water transfers the heat to the air at the radiator and fan. The cooling water circulates in the system through the water pump and is collected by the expansion tank.

[0106] The fuel cell system satisfies the energy balance equation:

[0107]

[0108] In the formula, C st M represents the specific heat capacity of the PEMFC stack. st T represents the mass of the fuel cell stack. st Q represents the temperature of the fuel cell stack. tot P represents the total chemical energy of the reactants. st Q represents the output power of the fuel cell stack. gas Q represents the net heat carried out of the system by the inlet and outlet gases. cl Q represents the heat carried out of the system by the cooling water. atm This represents the heat radiated from the system to the environment. Each term is calculated using the following equation:

[0109] P st =IV st

[0110] Q gas =Q in -Q out

[0111] Q cl =W cl C cl (T st -T in)

[0112]

[0113] The function of the expansion tank is to provide cooling water and, as heat capacity, reduce temperature fluctuations in the cooling water. It satisfies the following heat balance equation:

[0114]

[0115] The radiator satisfies the energy balance equation:

[0116]

[0117] The heat exchange between the radiator and the outside environment follows Newton's law of cooling:

[0118] Q rad =A rad (T rad -T atm )k rad

[0119] Heat exchanger average temperature T rad The average value of the radiator's inlet and outlet temperatures is taken.

[0120] In step S3, within the thermal management subsystem, the heat generated by the PEMFC system is first transferred to the cooling water and then to the atmosphere. Since the heat dissipation coefficient of forced air convection is much smaller than that of cooling water, more air is needed to remove the same amount of heat, resulting in the fan's power consumption being much higher than that of the water pump. To reduce the parasitic power of the PEMFC system, it is necessary to study the fan's power consumption. This paper establishes a power model for the fan based on its characteristic curves, as shown in Figure [Figure number missing]. Figure 3 As shown. The power of the fan can be expressed as:

[0121]

[0122] Where P s For the static pressure of the fan, η fan Where N is the fan efficiency, N is the rotational speed, and N0 is the maximum rotational speed. s and η fan It is a function of air velocity and can be obtained using polynomial fitting. Figure 3 P in s and η fan curve.

[0123] Step S4 requires establishing two PID controllers to control the inlet and outlet temperatures of the cooling water respectively. PID controllers are widely used in industrial process control due to their simple structure, good robustness, and independence from precise mathematical models of the system. The PID controller directly generates the corresponding control action based on the deviation between the given value and the actual system output.

[0124]

[0125] Where Kp is the proportional coefficient, Ki is the integral coefficient, Kd is the differential coefficient, e(t) represents the error between the actual temperature of the PEMFC and the given temperature, and t represents time;

[0126] The PEMFC thermal management system of this invention employs two independent PID controllers to control the speeds of the fan and water pump, respectively. For example... Figure 2 As shown, the inputs to the controller are the outlet and inlet water temperatures of the PEMFC, the control target is the given values ​​of the outlet and inlet water temperatures, and the output control action is the speed of the fan and water pump. Changes in the speed of the fan and water pump will cause changes in the cooling air flow and cooling water flow, thereby changing the heat exchange between the cooling water and air, as well as between the PEMFC battery stack and the cooling water, ultimately achieving the temperature control target.

[0127] Due to the nonlinear, strongly coupled, and long-delay characteristics of the PEMFC thermal management system, conventional PID control cannot achieve satisfactory control results. Therefore, some researchers have introduced fuzzy control to optimize conventional PID control, overcoming the limitations of fixed control parameters, and enabling online tuning of Kp, Ki, and Kd during the control process to better adapt to changes in the controlled object. The selection of membership functions and the formulation of fuzzy rules in fuzzy control have a significant impact on the control effect, but these parameters can only be obtained based on expert and engineering experience. Therefore, fuzzy control often fails to achieve satisfactory control results. In this context, this invention employs a reinforcement learning-based PID controller to achieve online tuning of PID control parameters. Comparison with fuzzy PID shows that it exhibits superior control performance and adaptive capability.

[0128] Step S5, reinforcement learning mainly consists of two parts: World (environment) and Agent (intelligent agent). When the agent interacts with the environment, the following events occur at each time t:

[0129] (1) The environmental state s(t) perceived by the agent at time t;

[0130] (2) Based on the current state s(t) and reinforcement learning information, the system selects and executes an action a(t);

[0131] (3) Action a(t) acts on the current environment, the environment changes, and enters a new state, i.e. s(t)→s(t+1);

[0132] (4) System feedback evaluation function;

[0133] (5) The agent accepts the evaluation function and changes the reinforcement information. The system returns to the previous step, i.e., s(t+1)→s(t);

[0134] (6) Return to step one and repeat the above steps until a satisfactory system state is obtained, then end the loop;

[0135] That is, the agent always chooses the action a(t) with the optimal evaluation function to obtain the maximum reward value of the environment (World);

[0136] like Figure 4 As shown, the reinforcement learning-based PID controller of the present invention consists of three parts: a performance evaluation unit, a parameter correction unit, and a decision unit.

[0137] The performance evaluation unit evaluates the current control effect and generates corresponding adjustment amounts based on the evaluation function;

[0138] The control effect evaluation function is defined as follows:

[0139] V(t) = exp(-|e(t)|)

[0140] The parameter adjustment algorithm is as follows:

[0141] ΔK p,t =α p (1-V(t))

[0142] ΔK i,t =α i (1-V(t))

[0143] ΔK dt =α d (1-V(t))

[0144] In the formula α p α i α d These are the learning rates of Kp, Ki, and Kd, respectively.

[0145] The parameter correction unit corrects Kp, Ki, and Kd according to the adjustment amount of the performance evaluation unit;

[0146] The correction algorithm is as follows:

[0147] K t+1 =K t

[0148]

[0149]

[0150]

[0151]

[0152] Where K represents Kp, Ki, and Kd, ΔK represents ΔKp, ΔKi, and ΔKd, the subscript t represents the current time, and t+1 represents the next time.

[0153] The decision-making unit finds a set of optimal values ​​from the adjustment amount, that is, obtains the adjustment amount with the optimal evaluation function value;

[0154] like Figure 4 As shown, the decision-making mechanism is that the PID controller generates five corresponding control actions U according to the updated five sets of parameters K, and applies them to the PEMFC thermal management system. The system enters the next time step t+1, the performance evaluation unit evaluates the control effect corresponding to the five control actions, the decision-making unit selects a set of parameters K with the optimal evaluation function, the system returns to the previous time step t and performs control according to the selected parameters K.

[0155] Reinforcement learning is a continuous "trial and error" process. In a reinforcement learning-based PID controller, the agent is a reinforcement learning PID, and the environment is a PEMFC thermal management system. First, all possible actions are generated through the performance evaluation unit. The parameter correction unit tries all actions and applies them to the current environment through the PID controller to obtain the evaluation function value. Finally, the decision unit selects the action with the optimal evaluation function value.

[0156] The invention will be further illustrated below through Example 1:

[0157] Example 1 is a PEMFC thermal management system with applied step current perturbation;

[0158] The step current disturbance, such as Figure 5 As shown;

[0159] Simulations were conducted based on this invention, and compared with existing PEMFC thermal management technologies. The experimental results applying this invention and results from other technologies are as follows:

[0160] from Figure 6It can be seen that, in terms of controlling the outlet cooling water temperature, the reinforcement learning PID of this invention has a better control effect than the fuzzy PID. Except that the overshoot is slightly greater than that of the fuzzy PID in the initial large error stage, the adjustment time is significantly shorter than that of the fuzzy PID. In the small error stage after stable operation, the reinforcement learning PID is better than the fuzzy PID in both overshoot and adjustment time.

[0161] from Figure 7 It can be seen that, apart from the significant changes in control parameters Kp, Ki, and Kd during the initial large error stage, the changes in Kp, Ki, and Kd are slight during the small error stage after stable operation, due to the small error and the small rate of change of the error.

[0162] from Figure 8 It can be seen that reinforcement learning PID can significantly change the control parameters Kp, Ki and Kd, whether in the initial large error stage or in the small error stage after stable operation, and its adaptive ability is stronger than that of fuzzy PID.

[0163] from Figure 9 It can be seen that the thermal management system of proton exchange membrane fuel cell has main and auxiliary heat sinks, which realizes the two-stage utilization of cooling air. Therefore, it can achieve the control target with less cooling air than the traditional PEMFC thermal management system, and at the same time, it can remove the fan from the inefficient zone and significantly reduce the fan parasitic power.

[0164] from Figure 10 It can be seen that, under the same control parameters, the thermal management system of proton exchange membrane fuel cell has a better control effect than the traditional PEMFC thermal management system, specifically in terms of smaller overshoot and shorter settling time.

[0165] The table below compares the control performance of the three controllers for the outlet cooling water temperature throughout the entire control process. IAE (Integral Absolute Error) is the absolute error integral criterion; a smaller IAE indicates better control performance. Its definition is as follows:

[0166] IAE=∫|e(t)|dt

[0167] controller PID Fuzzy PID Reinforcement learning PID outlet cooling water temperature IAE 347.7 268.9 192.1

[0168] It can be observed that reinforcement learning PID has the best control effect;

[0169] The table below compares the control performance of the proton exchange membrane fuel cell thermal management system and the traditional PEMFC thermal management system on the inlet cooling water temperature throughout the entire control process:

[0170]

[0171] It can be found that the thermal management system of the proton exchange membrane fuel cell reduces the inlet cooling water temperature (IAE) by 52.88% and the fan parasitic power by 69.94%. The improved thermal management system shows significant improvements in both control performance and parasitic power.

[0172] In summary, this invention provides a thermal management system for a proton exchange membrane fuel cell. By setting up main and auxiliary radiators, the cooling air blown in by the fan is reused, reducing the cooling airflow and allowing the fan to move out of the inefficient zone. This significantly reduces the parasitic work of the fan and improves the control effect of the inlet cooling water temperature. At the same time, this invention provides a reinforcement learning-based PID controller to optimize the PID on the water pump side, realizing adaptive changes in Kp, Ki, and Kd, thereby improving the control effect of the outlet cooling water temperature.

[0173] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or effective techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the appended claims.

Claims

1. A thermal management system for a proton exchange membrane fuel cell, characterized by, The thermal management system of the proton exchange membrane fuel cell is referred to as a PEMFC system, comprising: a fuel cell stack for realizing conversion of chemical energy in hydrogen into electric energy; a cooling water circulation loop, the cooling water circulation loop comprising a water pump, a heater, an expansion water tank, a main radiator and a secondary radiator; an output end of the water pump is connected to a cooling water inlet of the fuel cell stack; a cooling water outlet of the fuel cell stack is connected to an inlet of the heater and an inlet of the secondary radiator through two output ends of a three-way valve respectively; outlets of the heater and the main radiator are connected to an input end of the water pump; the cooling water circulation loop is divided into a heating branch and a heat dissipation branch; the heating branch is used for preheating the cooling water in a start-up stage of the PEMFC system; the heat dissipation branch is used for transferring heat generated by the fuel cell stack in a normal operation stage of the PEMFC system, controlling temperatures of the cooling water inlets and outlets, and maintaining efficient and stable operation of the system; the heat dissipation branch comprises a PEMFC, a particle filter, a three-way valve, a main radiator, a secondary radiator, an expansion water tank and a water pump, wherein the expansion water tank is used for providing the cooling water and plays a role of stabilizing pressure and temperature; the main radiator and the secondary radiator are connected by a closed channel and share a fan for cooling; after the cooling water flows through the particle filter, the cooling water is guided to the secondary radiator through the three-way valve to dissipate heat, then flows through the main radiator to further dissipate heat, and then flows to the fuel cell stack through the water pump; after the cooling water flows through the fuel cell stack, the cooling water flows through the particle filter again to perform a new cycle; a control module is used for controlling the three-way valve to be connected with the heating branch and disconnected with the heat dissipation branch when the PEMFC system is in the start-up stage, turning on the heater to preheat the cooling water to a given temperature; when the PEMFC system is in the normal operation stage, the three-way valve is controlled to be connected with the heat dissipation branch and disconnected with the heating branch, the heater is turned off, and two independent PID controllers are used to control the outlet temperature and the inlet temperature of the cooling water respectively.

2. A thermal management system for a proton exchange membrane fuel cell according to claim 1, wherein the heating branch comprises a fuel cell stack, a particle filter, a three-way valve, a heater, an expansion water tank and a water pump, wherein the expansion water tank is used for providing the cooling water and plays a role of stabilizing pressure and temperature; after the cooling water flows through the particle filter, the cooling water is guided to the heater through the three-way valve to be heated, and then flows to the fuel cell stack through the water pump; after the cooling water flows through the fuel cell stack, the cooling water flows through the particle filter again to perform a new cycle.

3. A thermal management system for a proton exchange membrane fuel cell according to claim 1, wherein In operation, the method comprises the following steps: in a start-up stage of the PEMFC system, the cooling water is preheated to a set temperature through the heating branch to avoid consumption of excessive hydrogen caused by the cooling water taking away heat of the fuel cell stack due to low temperature of the cooling water; in a normal operation stage of the PEMFC system, the cooling water flows through the heat dissipation branch to dissipate heat of the system, since a large amount of heat is generated by the fuel cell stack in the operation process and the inlet and outlet temperatures of the cooling water are controlled.

4. A temperature control method of a proton exchange membrane fuel cell, characterized by, The method is based on the thermal management system of the proton exchange membrane fuel cell of claim 1, comprising the following steps: S1: establishing a semi-empirical voltage model of the PEMFC to obtain an output electric power of the fuel cell stack; S2: Based on the output power of the battery stack obtained in S1, the physical characteristics of the water pump, radiator and expansion tank, the PEMFC thermal management system model is constructed according to the energy conservation equation, and is built based on the Matlab / Simulink platform to obtain the temperature change of the PEMFC system; S3: According to the fan characteristic curve, the fan power model is constructed, and the fan parasitic power problem is studied, which is used to calculate the control energy consumption of the controller established in S4 in the control process; S4: Based on the PEMFC thermal management system model established in S2, two independent PID controllers are established to adjust the speed of the fan and the water pump to control the inlet and outlet temperatures of the cooling water; S5: using reinforcement learning method to optimize the water pump side PID controller established in S4, to realize 、 and adaptive change in the control process.

5. A method of temperature control of a proton exchange membrane fuel cell as defined in claim 4, characterized in that In the step S1, the output power of the fuel cell stack is calculated by the following equation: wherein, represents the voltage loss due to the activation of electrochemical reactions, represents the voltage loss due to the activation of electrochemical reactions, represents the voltage loss due to the conduction of ions and electrons, represents the voltage loss due to the diffusion of reactant concentrations, represents the voltage of a monolithic cell, represents the voltage of a cell stack, represents the output electric power of a cell stack, N is the number of cell pieces.

6. The temperature control method of a proton exchange membrane fuel cell according to claim 4, characterized by, In the step S2, the energy balance equation for constructing the PEMFC thermal management system is as follows: wherein, represents the specific heat capacity of the PEMFC stack, represents the mass of the stack, represents the temperature of the stack, represents the total chemical energy of the reactants, represents the output power of the stack, represents the heat removed by the inlet and outlet gas purge system, represents the heat removed by the cooling water purge system, represents the heat radiated to the environment by the system.

7. The temperature control method of a proton exchange membrane fuel cell according to claim 4, characterized by, In the step S3, the fan power in the fan power model is calculated by the following equation: In the formula, Represents the actual fan speed. This represents the maximum fan speed. Represents airflow. Represents the static pressure of the fan. This represents the efficiency of the fan. and All are airflow The function is obtained by curve fitting in the fan characteristic curve. and The equation of the curve.

8. The temperature control method of a proton exchange membrane fuel cell according to claim 4, characterized by, In the step S4, the outlet cooling water temperature is controlled by adjusting the water pump speed, and the inlet cooling water temperature is controlled by adjusting the fan speed, and the PID controller satisfies the following equation in the time domain: wherein, is a proportional coefficient, is an integral coefficient, is a derivative coefficient, error function denotes the difference between the real-time temperature of the PEMFC and the target temperature.

9. The temperature control method of a proton exchange membrane fuel cell according to claim 4, characterized by, In the step S5, the reinforcement learning method adopted includes two parts: environment and agent, and when the agent interacts with the environment, the following events occur at each time t: (1) The agent perceives the state s(t) of the environment at time t; (2) According to the current state s(t) and reinforcement learning information, the system selects and executes a certain action a(t); (3) The action a(t) acts on the current environment, and the environment changes to a new state, i.e. s(t) → s(t+1); (4) The system feedback evaluation function; (5) The agent receives the evaluation function and changes the reinforcement information, and the system returns to the previous step, i.e. s(t+1) → s(t); (6) Return to step (1) and continue to repeat the above steps until the desired system state is obtained, and the cycle is ended; The PID controller based on the reinforcement learning method includes a performance evaluation unit, a parameter correction unit and a decision unit; The performance evaluation unit evaluates the current control effect, and generates a corresponding adjustment amount according to the evaluation function; The parameter correction unit corrects Kp, Ki and Kd according to the adjustment amount of the performance evaluation unit; The decision unit finds a set of optimal values from the adjustment amount, i.e. obtains the adjustment amount with the optimal evaluation function value.

Images