Hierarchical fuel cell temperature control method based on adaptive model-free predictive control

Abstract

The application discloses a layered fuel cell temperature control method based on adaptive model-free predictive control. First, the upper layer decision maker determines the optimal stack temperature under different working conditions based on the polarization curve of the stack, and outputs the optimal stack temperature as the optimal reference temperature to the lower layer controller. In the lower layer controller, the nonlinear prediction of the stack temperature is realized by introducing auxiliary relaxation variables based on the real-time input and output data of the battery operation. Then, in each execution cycle, the optimal control variable is obtained by solving the constrained optimization problem, to drive the water pump and radiator to complete the optimal control of the stack temperature. Compared with the traditional control method, the application reduces the dependence on the model; compared with the traditional model predictive control method, the application has the advantages of low calculation complexity, low requirement on the controller hardware, and the like, and to a certain extent, solves the problems of strong nonlinearity, strong coupling and great control difficulty of the proton exchange membrane fuel cell.

Description

Technical Field

[0001] This invention relates to a data-driven predictive control method in the field of battery thermal management, specifically to a control method and system for a layered proton exchange membrane fuel cell (PEMFC) thermal management system based on Adaptive Model-Free Predictive Control (AMFPC). Background Technology

[0002] Among various types of fuel cells, PEMFCs offer advantages such as high power density, rapid start-up capability, and suitability for both stationary and mobile applications. During operation, a PEMFC system generates electricity through the reaction of oxygen and hydrogen, producing only water and heat, ideally without any polluting waste. This makes it a key technology for the development and application of hydrogen energy in the context of green and sustainable development.

[0003] In fuel cell systems, stack temperature is one of the key parameters affecting fuel cell operating efficiency and system lifespan. Stack temperature directly impacts the output performance and durability of the PEMFC. When the stack temperature is too low, catalyst activity is inhibited, the electrochemical reaction rate decreases, and consequently, the fuel cell's output efficiency is significantly reduced. This not only affects the system's instantaneous output power but can also lead to a decrease in overall efficiency, especially during prolonged operation at lower temperatures, where the system's energy utilization rate is drastically reduced. Furthermore, low-temperature operation can easily lead to water management imbalances, causing moisture accumulation inside the stack, further affecting fuel cell performance. Conversely, when the stack temperature is too high, the problems are equally serious. High temperatures accelerate electrolyte membrane degradation, leading to decreased proton conductivity and even permanent membrane damage. In this case, not only does the stack's output performance decrease, but the lifespan of the entire fuel cell system is also greatly shortened, potentially even causing system failure. Therefore, precise control of the stack temperature is crucial in the design and operation of PEMFCs.

[0004] To ensure the fuel cell stack operates within its optimal temperature range, PEMFCs are typically equipped with a temperature management system. This system relies on two main actuators—a cooling water pump and a radiator (or fan)—to regulate the stack temperature. The cooling water pump controls system heat dissipation by adjusting the flow rate of cooling water into the stack. By adjusting the pump flow rate, heat generated by the stack can be promptly removed, preventing overheating. Secondly, the radiator is connected to the stack via piping, and the fan dissipates the heat removed from the stack into the air, ensuring the stack temperature does not exceed a safe range. The core principle of this temperature management system is to coordinate the cooling water flow rate and radiator efficiency through a reasonable control strategy to ensure that the heat generated by the chemical reactions inside the stack is dissipated in a timely and effective manner, thereby maintaining a stable stack temperature. This not only improves the immediate output performance of the fuel cell but also extends the overall lifespan of the system and enhances the long-term reliability of the fuel cell. Therefore, the effectiveness of PEMFC temperature management largely determines the system's operating efficiency and lifespan, making it a crucial aspect of fuel cell technology.

[0005] Currently, most control strategies for fuel cell stack temperature control employ model-based approaches, such as common feedforward, feedback, and model predictive control. However, the PEMFC thermal management system contains numerous nonlinear dynamics and semi-empirical equations to be fitted, making it difficult to obtain system equations with high fitting accuracy. This poses a challenge to model-based control strategies. Compared to learning-based model-free algorithms or traditional PID-based model-free control methods, the former incurs a significant computational burden, making direct implementation and widespread adoption difficult with current hardware conditions; the latter struggles to directly handle strongly coupled control systems with multiple inputs and outputs and suffers from complex parameter tuning issues.

[0006] Therefore, precise temperature control of fuel cells remains a significant challenge. Summary of the Invention

[0007] In view of this, in order to solve the problems existing in the background art, the purpose of this invention is to propose a hierarchical fuel cell thermal management system control method based on adaptive model-free predictive control.

[0008] The technical solution adopted in this invention is as follows:

[0009] I. A Layered Fuel Cell Temperature Control Method Based on Adaptive Model-Free Predictive Control

[0010] Step 1: Collect the actual stack temperature and actual control input of the stratified fuel cell, and generate the optimal stack temperature based on the current actual stack temperature;

[0011] Step 2: Construct a prediction model for the hierarchical fuel cell, and then establish an optimization objective for the hierarchical fuel cell based on the prediction model.

[0012] Step 3: Collect historical stack temperature and corresponding control input data of the stratified fuel cell. After periodically processing the historical stack temperature and corresponding control input data, obtain historical relative data and future relative data of the fuel cell. Then, combine the actual stack temperature and actual control input of the stratified fuel cell with the optimal stack temperature. Based on the optimization objective of the stratified fuel cell, optimize the cell temperature of the stratified fuel cell to obtain the optimal control sequence. Finally, control the thermal management system of the stratified fuel cell according to the optimal control sequence to complete the temperature control of the stratified fuel cell.

[0013] In step 1, the optimal stack temperature is generated based on the current actual stack temperature, specifically as follows:

[0014] Find the stack temperature Tr corresponding to the maximum output power point on the voltage-current polarization curve of the stack at the current actual stack temperature, and take this stack temperature Tr as the optimal stack temperature for the current actual stack temperature.

[0015] In step 2, the optimization objective of the stratified fuel cell is formulated as follows:

[0016]

[0017] Where, σ y Let g represent the slack variable, u represent the predictive control input of the thermal management system, y represent the predicted stack temperature, and N represent the prediction step size. k r represents the predicted stack temperature at the k-th step. t+k U represents the reference temperature of the fuel cell stack at time t+k. k This represents the control input for the k-th step, and the first optimization term. Satisfy (||y) k -r t+k || T *Q*||y k -r t+k ||) 2 The second optimization term Satisfy (||u k || T *R*||u k ||) 2 |||1 denotes the 1-norm, T denotes the transpose, R and Q are the first and second coefficient matrices respectively, ||| denotes a matrix, and λ g and λ y There are two penalty coefficients. This is the set of constraints for the actual stack temperature. The set of constraints for controlling the input; U pY represents the historical relative control input matrix in the historical relative data of fuel cells. p U represents the historical relative stack temperature matrix in the historical relative data of fuel cells. f Y represents the future relative control input matrix in the future relative data of the fuel cell. f This represents the future relative stack temperature matrix in the future relative data of the fuel cell, u ini y represents the actual control input of the stratified fuel cell. ini This indicates the actual stack temperature of the stratified fuel cell.

[0018] In step 3, after periodically processing the historical stack temperature and control input data, historical relative data and future relative data of the fuel cell are obtained, specifically as follows:

[0019] After processing the historical stack temperature and control input data using the following formula, the historical relative control input matrix U is obtained. p The historical relative stack temperature matrix Y p This data forms the historical relative data of the fuel cell and provides the future relative control input matrix U. f and the future relative stack temperature matrix Y f And form the relative data for the future of fuel cells:

[0020]

[0021] Where L represents the proportionality coefficient. Represents the Hankel matrix, u d It is the collected historical control input data, y d For the historical fuel cell stack temperature data collected, and These are historical control input data u d The first one, the TT s +1, the Lth and the Tth s Each component, y1, y2 T-L+1 y L and y T These are historical fuel cell stack temperature data y d The first one, the TT s +1, the Lth and Tth s One component, T s Represents historical fuel cell stack temperature data u d Data length, T ini represents the initial time, and N represents the prediction step size.

[0022] II. A hierarchical fuel cell temperature control system based on adaptive model-free predictive control

[0023] The upper-level decision-maker is used to generate the optimal stack temperature based on the actual stack temperature of the stratified fuel cell collected.

[0024] The lower-level controller is used to collect historical stack temperatures and corresponding control input data of the stratified fuel cell. After periodically processing the historical stack temperatures and corresponding control input data, it obtains historical relative data and future relative data of the fuel cell. Then, combined with the actual stack temperature and actual control input of the stratified fuel cell, and the optimal stack temperature, it optimizes the cell temperature of the stratified fuel cell according to the optimization objective of the stratified fuel cell, obtains the optimal control sequence, and applies it to the thermal management system of the stratified fuel cell.

[0025] III. A computer device

[0026] The computer device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the hierarchical fuel cell temperature control method based on adaptive model-free predictive control.

[0027] IV. A computer-readable storage medium

[0028] The storage medium stores a computer program, which, when executed by a processor, implements the steps of any of the described hierarchical fuel cell temperature control methods based on adaptive model-free predictive control.

[0029] V. A computer program product

[0030] The product includes a computer program / instructions that, when executed by a processor, implement the steps of the hierarchical fuel cell temperature control method based on adaptive model-free predictive control.

[0031] The beneficial effects of this invention are as follows:

[0032] (1) This invention combines fuel cell thermal management model and operating data, predicts output through decision variables, and introduces slack variables for nonlinear compensation, which overcomes the shortcomings of traditional battery temperature control methods that are limited by strong system coupling and strong nonlinearity. At the same time, compared with learning algorithms (based on neural networks, deep learning, reinforcement learning, etc.), it greatly reduces the computational burden and improves the stability of fuel cell operating temperature.

[0033] (2) This invention proposes a hierarchical optimal temperature control framework to achieve optimal temperature dynamic tracking effect under different operating conditions and maximize the output power of the fuel cell stack.

[0034] Therefore, this invention reduces the dependence on models compared to traditional control methods; compared to traditional model predictive control methods, this invention has advantages such as low computational complexity and low requirements for controller hardware, and to a certain extent solves the problems of strong nonlinearity, strong coupling and high control difficulty of proton exchange membrane fuel cells. Attached Figure Description

[0035] Figure 1 This is a framework diagram of the hierarchical control method involved in the present invention.

[0036] Figure 2 This is a structural diagram of the AMFPC control method involved in the present invention.

[0037] Figure 3 This is a diagram illustrating the effectiveness verification of the AMFPC control method involved in this invention. Detailed Implementation

[0038] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited thereto.

[0039] For fuel cell systems characterized by strong coupling, nonlinearity, and multiple inputs and outputs, this invention proposes a hierarchical fuel cell temperature control method based on adaptive model-free predictive control. This method employs a hierarchical control framework, such as... Figure 1 As shown, the hierarchical control framework includes an upper-level decision-maker and a lower-level controller. The upper-level decision-maker determines the optimal stack temperature based on the stack polarization curve. The lower-level controller relies on traditional PID control for brief data acquisition during the data acquisition phase, and then quickly switches to the AMFPC method to achieve real-time online tracking of the optimal temperature. The circulating water pump and cooling fan of the PEMFC temperature management system are used as actuators, and the stack output temperature is the control variable.

[0040] This embodiment specifically includes the following steps:

[0041] Step 1: Collect the actual stack temperature and actual control input of the stratified fuel cell, and generate the optimal stack temperature based on the current actual stack temperature, which is used as the reference stack temperature;

[0042] Specifically, the optimal stack temperature is generated based on the current actual stack temperature, as follows:

[0043] like Figure 1 As shown, the stack temperature Tr corresponding to the maximum output power point is found in the voltage-current polarization curve of the stack corresponding to the current actual stack temperature, and this stack temperature Tr is taken as the optimal stack temperature under the current actual stack temperature (i.e., under this operating condition).

[0044] Step 2: Construct a predictive model for the stratified fuel cell, and then establish an optimization objective for the stratified fuel cell based on the predictive model, such as... Figure 2 As shown;

[0045] The prediction model for stratified fuel cells is an adaptive model-free prediction model, and its formula is as follows:

[0046]

[0047] To achieve the minimum error and minimize control action costs, and considering the constraints of actual industrial conditions, as well as constraints on slack variables and decision variables, the optimization objective of the hierarchical fuel cell is formulated as follows:

[0048]

[0049] Where, σ y Let g represent the slack variable, u represent the predictive control input of the thermal management system, y represent the predicted stack temperature, and N represent the prediction step size. k r represents the predicted stack temperature at the k-th step. t+k U represents the reference temperature of the fuel cell stack at time t+k. k This represents the control input for the k-th step, and the first optimization term. Satisfy (||y) k -r t+k || T *Q*||y k -r t+k ||) 2 The second optimization term Satisfy (||u k || T *R*||u k ||) 2 |||1 denotes the 1-norm, T denotes the transpose, R and Q are the first and second coefficient matrices respectively, ||| denotes a matrix, and λ g and λ y There are two penalty coefficients. This is the set of constraints for the actual stack temperature. The set of constraints for controlling the input; U p Y represents the historical relative control input matrix in the historical relative data of fuel cells. p U represents the historical relative stack temperature matrix in the historical relative data of fuel cells. f Y represents the future relative control input matrix in the future relative data of the fuel cell. f This represents the future relative stack temperature matrix in the future relative data of the fuel cell, u ini y represents the actual control input of the stratified fuel cell. ini This indicates the actual stack temperature of the stratified fuel cell.

[0050] Step 3: Before collecting historical stack temperature and corresponding control input data of the stratified fuel cell, the controller switches to the classic fuzzy PID control method. Then, it collects the historical stack temperature and corresponding control input data. After data collection is complete, it switches to the model-free predictive control method proposed in this invention. Next, the historical stack temperature and corresponding control input data are periodically processed to obtain historical relative data and future relative data of the fuel cell. Then, combined with the actual stack temperature and actual control input of the stratified fuel cell, and the optimal stack temperature, the cell temperature of the stratified fuel cell is optimized according to the optimization objective of the stratified fuel cell to obtain the optimal control sequence. Figure 1 The lower-level controller in the process ultimately controls the thermal management system of the stratified fuel cell according to the optimal control sequence. Specifically, the first control variable of the optimal control sequence is applied to the thermal management system to complete the temperature control of the stratified fuel cell.

[0051] In step 3, after periodically processing the historical stack temperature and control input data, historical relative data and future relative data of the fuel cell are obtained, specifically:

[0052] After processing the historical stack temperature and control input data using the following formula, the historical relative control input matrix U is obtained. p The historical relative stack temperature matrix Y p This data forms the historical relative data of the fuel cell and provides the future relative control input matrix U. f and the future relative stack temperature matrix Y f And form the relative data for the future of fuel cells:

[0053]

[0054] Where L represents the proportionality coefficient. Represents the Hankel matrix, u d It is the collected historical control input data, y d To collect historical fuel cell stack temperature data, u d and y d Both are vectors. and These are historical control input data u d The first one, the TT s +1, the Lth and the Tth s Each component, y1, y2 T-L+1 y L and y T These are historical fuel cell stack temperature data y d The first one, the TT s +1, the Lth and Tth sOne component, T s Represents historical fuel cell stack temperature data u d and historical fuel cell stack temperature data y d Data length, T ini represents the initial time, and N represents the prediction step size.

[0055] This invention also proposes a hierarchical fuel cell temperature control system based on adaptive model-free predictive control, comprising:

[0056] The upper-level decision-maker is used to generate the optimal stack temperature based on the actual stack temperature of the stratified fuel cell collected.

[0057] The lower-level controller (i.e., the AMFPC controller) is used to collect historical stack temperatures and corresponding control input data of the stratified fuel cell. After periodically processing the historical stack temperatures and corresponding control input data, it obtains historical relative data and future relative data of the fuel cell. Then, combined with the actual stack temperature and actual control input of the stratified fuel cell, and the optimal stack temperature, it optimizes the cell temperature of the stratified fuel cell according to the optimization objective of the stratified fuel cell, obtains the optimal control sequence, and applies it to the thermal management system of the stratified fuel cell.

[0058] The AMFPC method proposed in this invention is verified below. With the initial temperature of the fuel cell set at 62°C, verification graphs of the control effect of this invention are generated for different optimal temperature requirements, as shown below. Figure 3 As shown in the figure, the AMFPC method designed in this invention can control the output temperature to reach the reference temperature at a relatively fast speed and maintain a stable state, with an absolute error of less than 0.5℃ and a dynamic error of less than 2℃. Therefore, it can be proven that the AMFPC method designed in this invention has good control performance.

[0059] Finally, it should be noted that the above embodiments and descriptions are only used to illustrate the technical solutions of the present invention and not to limit it. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the disclosure of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the protection scope of the claims of the present invention.

Claims

1. A hierarchical fuel cell temperature control method based on adaptive model-free predictive control, characterized in that, Includes the following steps: Step 1: Collect the actual stack temperature and actual control input of the stratified fuel cell, and generate the optimal stack temperature based on the current actual stack temperature; Step 2: Construct a prediction model for the hierarchical fuel cell, and then establish an optimization objective for the hierarchical fuel cell based on the prediction model. Step 3: Collect historical stack temperature and corresponding control input data of the stratified fuel cell. After periodically processing the historical stack temperature and corresponding control input data, obtain historical relative data and future relative data of the fuel cell. Then, combine the actual stack temperature and actual control input of the stratified fuel cell with the optimal stack temperature. Based on the optimization objective of the stratified fuel cell, optimize the cell temperature of the stratified fuel cell to obtain the optimal control sequence. Finally, control the thermal management system of the stratified fuel cell according to the optimal control sequence to complete the temperature control of the stratified fuel cell. In step 2, the optimization objective of the stratified fuel cell is formulated as follows: in, Let g represent the slack variable, u represent the predictive control input of the thermal management system, y represent the predicted stack temperature, and N represent the prediction step size. k r represents the predicted stack temperature at the k-th step. t+k U represents the reference temperature of the fuel cell stack at time t+k. k This represents the control input for the k-th step, and the first optimization term. satisfy The second optimization term satisfy , Let T denote the norm, T denote the transpose, and R and Q be the first and second coefficient matrices, respectively. Represents a matrix, and There are two penalty coefficients. This is the set of constraints for the actual stack temperature. The set of constraints for controlling the input; U p Y represents the historical relative control input matrix in the historical relative data of fuel cells. p U represents the historical relative stack temperature matrix in the historical relative data of fuel cells. f Y represents the future relative control input matrix in the future relative data of the fuel cell. f This represents the future relative stack temperature matrix in the future relative data of the fuel cell, u ini y represents the actual control input of the stratified fuel cell. ini This indicates the actual stack temperature of the stratified fuel cell; In step 3, after periodically processing the historical stack temperature and control input data, historical relative data and future relative data of the fuel cell are obtained, specifically as follows: After processing the historical stack temperature and control input data using the following formula, the historical relative control input matrix U is obtained. p The historical relative stack temperature matrix Y p This data forms the historical relative data of the fuel cell and provides the future relative control input matrix U. f and the future relative stack temperature matrix Y f And form the relative data for the future of fuel cells: in, Represents the proportionality coefficient. Represents the Hankel matrix. It is the collected historical control input data. For the historical fuel cell stack temperature data collected, , , and These are historical control input data u d The first one, the Tth s -L+1, the Lth and the Tth s One portion, , , and These are historical fuel cell stack temperature data y d The first one, the Tth s -L+1, the Lth and Tth s One component, T s y represents historical fuel cell stack temperature data d Data length, represents the initial time, and N represents the prediction step size.

2. The hierarchical fuel cell temperature control method based on adaptive model-free predictive control according to claim 1, characterized in that, In step 1, the optimal stack temperature is generated based on the current actual stack temperature, specifically as follows: Find the stack temperature Tr corresponding to the maximum output power point on the voltage-current polarization curve of the stack at the current actual stack temperature, and take this stack temperature Tr as the optimal stack temperature for the current actual stack temperature.

3. A hierarchical fuel cell temperature control system based on adaptive model-free predictive control for implementing the method of claim 1, characterized in that, include: The upper-level decision-maker is used to generate the optimal stack temperature based on the actual stack temperature of the stratified fuel cell collected. The lower-level controller is used to collect historical stack temperatures and corresponding control input data of the stratified fuel cell. After periodically processing the historical stack temperatures and corresponding control input data, it obtains historical relative data and future relative data of the fuel cell. Then, combined with the actual stack temperature and actual control input of the stratified fuel cell, and the optimal stack temperature, it optimizes the cell temperature of the stratified fuel cell according to the optimization objective of the stratified fuel cell, obtains the optimal control sequence, and applies it to the thermal management system of the stratified fuel cell.

4. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the hierarchical fuel cell temperature control method based on adaptive model-free predictive control as described in any one of claims 1 to 2.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the hierarchical fuel cell temperature control method based on adaptive model-free predictive control as described in any one of claims 1 to 2.

6. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the hierarchical fuel cell temperature control method based on adaptive model-free predictive control as described in any one of claims 1 to 2.

Images