Method for controlling intelligent health management of fuel cell

Abstract

Provided in the present invention is a method for controlling intelligent health management of a fuel cell. The method comprises: taking, as control target values of a fuel cell, a current, a temperature, a hydrogen emission, an oxygen emission, a hydrogen reflux quantity, a hydrogen-side pressure and an oxygen-side pressure, which correspond to each of a calculated value of the total voltage of a fuel cell stack, a calculated value of a single-cell voltage mean square error of the fuel cell stack, a calculated value of a single-cell voltage change rate of the fuel cell stack and a calculated value of a single-cell voltage minimum amplitude of the fuel cell stack, or correspond to each of an optimized value of the total voltage of the fuel cell stack, an optimized value of the single-cell voltage mean square error of the fuel cell stack, an optimized value of the single-cell voltage change rate of the fuel cell stack and an optimized value of the single-cell voltage minimum amplitude of the fuel cell stack, and thus a membrane electrode is always in a "water balance" state, thereby improving the performance of cells, reducing the occurrence of cell faults, and improving the self-adaptability, reliability and stability of a fuel cell system.

Description

Intelligent health management control method for fuel cells Technical Field

[0001] The present invention relates to the field of energy technology, and more specifically, to a fuel cell intelligent health management and control method. Background Art

[0002] Proton exchange membrane fuel cells (PEMFCs) use hydrogen to react with oxygen in the air or pure oxygen to directly convert chemical energy into electrical energy. They are characterized by high energy conversion efficiency, low operating noise, high reliability, low operating temperatures, low-temperature startup, zero emissions, environmental friendliness, simple structure, and easy operation. Hydrogen is a widely available, renewable energy resource, making PEMFCs a truly green and environmentally friendly energy source that is increasingly being applied in various new energy fields.

[0003] Fuel cells conduct reactions through membrane electrodes. The polymer electrolyte in the membrane electrode must contain water during operation to ensure proton conductivity. However, this water content cannot be too high, otherwise it will clog the gas diffusion layer of the membrane electrode and affect the reaction. Therefore, it is necessary to maintain a dynamic balance of water content in the membrane electrode. In actual operation, many factors affect the water content of the membrane electrode. Especially at the system level, when multiple single cells are connected in series or even multiple stacks are used in series and parallel, the control of the water content of each membrane electrode section is even more difficult.

[0004] Patent CN112117475A uses a learning vector quantization neural network algorithm to detect faults in the water management subsystem. The training samples use parameters such as the pressure and temperature of the anode and cathode of the system. Patent CN114976133A uses a model to establish a fault-tolerant control method for fuel cell water management. It designs a flooding fault factor based on a water transmission model, and then uses this to design a fault-tolerant controller to adjust the cathode air intake to control flooding faults. In the prior art, a model is designed for a single target such as air flow, air pressure, and stack temperature, and a linear relationship or PID control model is established. This lacks overall consideration. In actual fuel cell systems, there are many factors that affect the water content of the membrane electrode, and it is the result of the combined action of these factors.

[0005] In view of this, the present invention provides a fuel cell intelligent health management control method.

[0006] Summary of the Invention

[0007] In response to the technical problems raised above, a method for intelligent fuel cell health management and control is provided. This invention primarily utilizes a recurrent neural network to establish a system model. Using a differential evolution algorithm with stack inspection parameters as the objective function, it determines an optimized set of input parameters to indirectly control the membrane electrode moisture content, thereby better simulating the operation of the fuel cell system. Furthermore, combined with an optimization algorithm, the optimized parameters are dynamically calculated to achieve better adaptive control.

[0008] The technical means adopted in the present invention are as follows:

[0009] The present invention provides a fuel cell intelligent health management control method, comprising:

[0010] Inputting the current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux, hydrogen side pressure, and oxygen side pressure of the fuel cell into the trained nonlinear mapping model, the trained nonlinear mapping model outputting a calculation result;

[0011] Input the calculation result into the objective function, and find the calculated value of the total voltage of the stack, the calculated value of the mean square deviation of the voltage of a single cell of the stack, the calculated value of the rate of change of the voltage of a single cell of the stack, and the calculated value of the minimum amplitude of the voltage of a single cell of the stack corresponding to the maximum value of the objective function;

[0012] Measure the actual value of the total voltage of the battery stack, the actual value of the mean square deviation of the voltage of a single cell of the battery stack, the actual value of the voltage change rate of a single cell of the battery stack, and the actual value of the minimum amplitude of the voltage of a single cell of the battery stack; the difference between the calculated value of the total voltage of the battery stack and the actual value of the total voltage of the battery stack is a first difference, the difference between the calculated value of the mean square deviation of the voltage of a single cell of the battery stack and the actual value of the mean square deviation of the voltage of a single cell of the battery stack is a second difference, the difference between the calculated value of the voltage change rate of a single cell of the battery stack and the actual value of the voltage change rate of a single cell of the battery stack is a third difference, and the difference between the calculated value of the minimum amplitude of the voltage of a single cell of the battery stack and the actual value of the minimum amplitude of the voltage of a single cell of the battery stack is a fourth difference;

[0013] respectively determining whether the first difference is greater than a first threshold, whether the second difference is greater than a second threshold, whether the third difference is greater than a third threshold, and whether the fourth difference is greater than a fourth threshold;

[0014] If the first difference is less than or equal to the first threshold, the second difference is less than or equal to the second threshold, the third difference is less than or equal to the third threshold, and the fourth difference is less than or equal to the fourth threshold, the current, the temperature, the hydrogen displacement, the oxygen displacement, the hydrogen reflux amount, the hydrogen side pressure and the oxygen side pressure corresponding to the calculated value of the total stack voltage, the calculated value of the mean square deviation of the stack single-cell voltage, the calculated value of the stack single-cell voltage change rate and the calculated value of the minimum amplitude of the stack single-cell voltage are used as the control target values ​​of the fuel cell.

[0015] Furthermore, the objective function is calculated as follows: Y=op1×A+(1 / op2)×B+(1 / op3)×C+(1 / op4)×D;

[0016] Among them, op1 is the calculated value of the total voltage of the battery stack, A is the first weight, op2 is the calculated value of the mean square deviation of the voltage of a single cell of the battery stack, B is the second weight, op3 is the calculated value of the voltage change rate of a single cell of the battery stack, C is the third weight, op4 is the calculated value of the minimum amplitude of the voltage of a single cell of the battery stack, and D is the fourth weight.

[0017] Further, if the first difference is greater than the first threshold, adjusting the first weight to a first optimized weight according to a fuzzy control algorithm;

[0018] If the second difference is greater than the second threshold, adjusting the second weight to a second optimized weight according to a fuzzy control algorithm;

[0019] If the third difference is greater than the third threshold, adjusting the third weight to a third optimized weight according to a fuzzy control algorithm;

[0020] If the fourth difference is greater than the fourth threshold, the fourth weight is adjusted to a fourth optimized weight according to a fuzzy control algorithm.

[0021] Further, when at least one of the first weight, the second weight, the third weight and the fourth weight of the objective function changes to form an optimized objective function;

[0022] Finding the optimized value of the total voltage of the stack, the optimized value of the mean square deviation of the voltage of a single cell of the stack, the optimized value of the rate of change of the voltage of a single cell of the stack, and the optimized value of the minimum amplitude of the voltage of a single cell of the stack, which correspond to the maximum value of the optimization objective function;

[0023] The current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux flow, hydrogen side pressure and oxygen side pressure corresponding to the optimized value of the total stack voltage, the optimized value of the mean square deviation of the single-cell stack voltage, the optimized value of the single-cell stack voltage change rate and the optimized value of the minimum amplitude of the single-cell stack voltage are used as the control target values ​​of the fuel cell.

[0024] Furthermore, the nonlinear mapping model includes an input layer, an intermediate layer and an output layer connected in sequence, the intermediate layer includes a plurality of neurons; and the intermediate layer is connected to a hidden layer.

[0025] Furthermore, training the nonlinear mapping model includes:

[0026] Provide datasets;

[0027] performing normalization processing on the data set to obtain a processed data set;

[0028] The data set is input into the nonlinear mapping model, and after a preset number of iterations, the trained nonlinear mapping model is obtained.

[0029] Furthermore, the weight update learning rate of the input layer is in the range of 0.2 to 0.3, the weight update learning rate of the output layer is in the range of 0.2 to 0.4, and the weight update learning rate of the hidden layer is in the range of 0.1 to 0.2.

[0030] Compared with the prior art, the present invention has the following advantages:

[0031] The intelligent fuel cell health management and control method provided by the present invention uses the current corresponding to the calculated value of the total stack voltage, the calculated value of the mean square error of the voltage of each cell, the calculated value of the voltage change rate of each cell, and the calculated value of the minimum voltage amplitude of each cell as targets, ensuring that the fuel cell operates under reasonable operating conditions and avoiding failures such as "dry membrane" or "flooding" of the fuel cell membrane electrode. When health deteriorates or even fails, the optimization target is adjusted and the operating conditions are changed, targeting the optimized values ​​of the total stack voltage, the mean square error of the voltage of each cell, the calculated value of the voltage change rate of each cell, and the calculated value of the minimum voltage amplitude of each cell, so that the membrane electrode reaches a new "water balance" state, thereby improving battery performance, reducing the occurrence of battery failures, and enhancing the adaptability, reliability, and stability of the fuel cell system. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0033] FIG1 is a flow chart of a fuel cell intelligent health management control method provided by the present invention.

[0034] FIG2 is another flow chart of the intelligent health management control method for a fuel cell provided by the present invention.

[0035] FIG3 is another flow chart of the intelligent health management control method for a fuel cell provided by the present invention. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0038] In conjunction with FIG1 and FIG2 , FIG1 is a flow chart of a fuel cell intelligent health management control method provided by the present invention, and FIG2 is another flow chart of a fuel cell intelligent health management control method provided by the present invention, to illustrate a specific embodiment of the fuel cell intelligent health management control method provided by the present invention, including:

[0039] S1: Input the fuel cell current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux, hydrogen side pressure, and oxygen side pressure into the trained nonlinear mapping model, and the trained nonlinear mapping model outputs the calculation results;

[0040] S2: Input the calculation results into the objective function to find the calculated value of the total stack voltage, the calculated value of the mean square error of the voltage of a single cell in the stack, the calculated value of the voltage change rate of a single cell in the stack, and the calculated value of the minimum amplitude of the voltage of a single cell in the stack corresponding to the maximum value of the objective function;

[0041] S3: Measure the actual value of the total voltage of the stack, the actual value of the mean square deviation of the voltage of a single cell of the stack, the actual value of the voltage change rate of a single cell of the stack, and the actual value of the minimum amplitude of the voltage of a single cell of the stack; the difference between the calculated value of the total voltage of the stack and the actual value of the total voltage of the stack is the first difference, the difference between the calculated value of the mean square deviation of the voltage of a single cell of the stack and the actual value of the mean square deviation of the voltage of a single cell of the stack is the second difference, the difference between the calculated value of the voltage change rate of a single cell of the stack and the actual value of the voltage change rate of a single cell of the stack is the third difference, and the difference between the calculated value of the minimum amplitude of the voltage of a single cell of the stack and the actual value of the minimum amplitude of the voltage of a single cell of the stack is the fourth difference;

[0042] S4: respectively determining whether the first difference is greater than a first threshold, whether the second difference is greater than a second threshold, whether the third difference is greater than a third threshold, and whether the fourth difference is greater than a fourth threshold;

[0043] S5: If the first difference is less than or equal to the first threshold, the second difference is less than or equal to the second threshold, the third difference is less than or equal to the third threshold, and the fourth difference is less than or equal to the fourth threshold, the current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux flow, hydrogen side pressure and oxygen side pressure corresponding to the calculated value of the total voltage of the stack, the calculated value of the mean square deviation of the voltage of a single cell of the stack, the calculated value of the voltage change rate of a single cell of the stack and the calculated value of the minimum amplitude of the voltage of a single cell of the stack are used as the control target values ​​of the fuel cell.

[0044] It is understood that the output calculation results of the trained nonlinear mapping model are used as the objective function of the intelligent optimization algorithm using a weighted summation method to guide the evolutionary algorithm in its search for a more optimal solution. The intelligent optimization algorithm uses a differential evolution algorithm. Based on the evaluation of the objective function, it iteratively calculates the input parameter combination that maximizes the objective function. The intelligent optimization algorithm is written in embedded C to run on a single-chip microcontroller controller. It calculates the input parameters that approximate optimal performance in real time and uses these input parameters as control parameters.

[0045] Compared with the prior art, the fuel cell intelligent health management control method provided by the present invention has at least the following advantages:

[0046] The intelligent fuel cell health management and control method provided by the present invention uses the current corresponding to the calculated value of the total stack voltage, the calculated value of the mean square error of the voltage of each cell, the calculated value of the voltage change rate of each cell, and the calculated value of the minimum voltage amplitude of each cell as targets, ensuring that the fuel cell operates under reasonable operating conditions and avoiding failures such as "dry membrane" or "flooding" of the fuel cell membrane electrode. When health deteriorates or even fails, the optimization target is adjusted and the operating conditions are changed, targeting the optimized values ​​of the total stack voltage, the mean square error of the voltage of each cell, the calculated value of the voltage change rate of each cell, and the calculated value of the minimum voltage amplitude of each cell, so that the membrane electrode reaches a new "water balance" state, thereby improving battery performance, reducing the occurrence of battery failures, and enhancing the adaptability, reliability, and stability of the fuel cell system.

[0047] In some optional embodiments, with continued reference to FIG. 1 and FIG. 2 , the objective function is calculated as follows: Y=op1×A+(1 / op2)×B+(1 / op3)×C+(1 / op4)×D;

[0048] Among them, op1 is the calculated value of the total voltage of the stack, A is the first weight, op2 is the calculated value of the mean square deviation of the voltage of a single cell of the stack, B is the second weight, op3 is the calculated value of the voltage change rate of a single cell of the stack, C is the third weight, op4 is the calculated value of the minimum amplitude of the voltage of a single cell of the stack, and D is the fourth weight.

[0049] It can be understood that the calculated value of the total stack voltage is the sum of the single-cell voltages of the stack inspection, the calculated value of the mean square deviation of the single-cell voltages of the stack inspection, the calculated value of the change rate of the single-cell voltage of the stack inspection is the amplitude change rate of the single-cell voltage of the stack inspection, and the calculated value of the minimum amplitude of the single-cell voltage of the stack inspection is the minimum amplitude of the single-cell voltage of the stack inspection. These four parameters are used to express the water content of the fuel cell membrane electrode.

[0050] The design of the differential evolution optimization algorithm prioritizes real-time performance while also taking into account the degree of optimization. Specifically, the algorithm limits the population size to 20, the individual length to 3, the mutation control parameter to 0.5, and the crossover control parameter to 0.8. This reduces computational effort and enables rapid algorithm convergence. The output of the trained nonlinear mapping model is weighted according to the optimization objective and serves as the final objective function for the differential evolution algorithm. The input parameters are not directly continuous random numbers, but rather randomly selected from a limited array designed based on empirical methods and the physical constraints of actual components. This approach sacrifices accuracy in finding the optimal solution, but achieves the more important benefit of real-time performance. The algorithm calculates near-optimal input parameters in real time based on known parameter values, and these input parameters serve as control inputs for the corresponding control components. In practical control systems, empirical control methods can also be incorporated to effectively mitigate the effects of system disturbances.

[0051] In some optional embodiments, continuing to refer to FIG1 and FIG2 , if the first difference is greater than the first threshold, the first weight is adjusted to a first optimized weight according to the fuzzy control algorithm;

[0052] If the second difference is greater than a second threshold, adjusting the second weight to a second optimized weight according to the fuzzy control algorithm;

[0053] If the third difference is greater than a third threshold, adjusting the third weight to a third optimized weight according to the fuzzy control algorithm;

[0054] If the fourth difference is greater than a fourth threshold, the fourth weight is adjusted to a fourth optimized weight according to the fuzzy control algorithm.

[0055] It is understandable that the specific adjustment methods are as follows:

[0056] The weight adjustment is calculated using a fuzzy control algorithm. Taking the calculated total voltage of the battery stack as an example, if the deviation between the calculated total voltage of the battery stack and the actual total voltage of the battery stack exceeds the first threshold, it is considered that an abnormal operating condition has occurred, and the first weight in the objective function is increased. In the design of the fuzzy control algorithm for adjusting the specific weights, the input design is the deviation between the optimization item and the design target value. The membership function is defined according to the severity of the deviation. The membership function is divided as follows:

[0057] Input membership: Deviation e = Vset-V, Vset is the calculated value of the total voltage of the fuel cell stack, and V is the actual value of the total voltage of the fuel cell stack.

[0058] VH interval: if e>12V else y=1; if e>9V and e<12else y=(e-9) / 3;

[0059] NH interval: if e>9V and e<12else y=(12-e) / 3; if e>6V and e<9else y=(e-6) / 3;

[0060] NM interval: if e>6V and e<9else y=(9-e) / 3; if e>3V and e<6else y=(e-3) / 3;

[0061] NL interval: if e>3V and e<6else y=(6-e) / 3; if e>0V and e<3else y=e / 3;

[0062] VL interval: if e>0V and e<3else y=(3-e) / 3; if e<0V else y=1;

[0063] Output membership:

[0064] VB interval: if e>2else y=1; if e>1.5and e<2else y=(e-1.5) / 0.5;

[0065] NB interval: if e>1.5and e<2else y=(2-e) / 0.5; if e>1and e<1.5else y=(e-1) / 0.5;

[0066] NS interval: if e>1and e<1.5else y=(1.5-e) / 0.5; if e>0.5and e<1else y=(e-0.5) / 0.5;

[0067] VS interval: if e>0.5and e<1else y=(1-e) / 0.5; if e<0.5else y=1;

[0068] Specifically, the fuzzy control rule base is shown in Table 1 as follows:

[0069] Table 1. Fuzzy control rule base

[0070] In some optional embodiments, with continued reference to FIG2 and FIG3 , when at least one of the first weight, the second weight, the third weight, and the fourth weight of the objective function changes to form an optimized objective function;

[0071] Find the optimal value of the total voltage of the stack, the optimal value of the mean square deviation of the voltage of a single cell in the stack, the optimal value of the rate of change of the voltage of a single cell in the stack, and the optimal value of the minimum amplitude of the voltage of a single cell in the stack, which correspond to the maximum value of the optimization objective function;

[0072] The current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux flow, hydrogen side pressure and oxygen side pressure corresponding to the optimized value of the total stack voltage, the optimized value of the mean square deviation of the single-cell stack voltage, the optimized value of the voltage change rate of the single-cell stack and the optimized value of the minimum amplitude of the single-cell stack voltage are used as the control target values ​​of the fuel cell.

[0073] It is understandable that the optimized values ​​of the total stack voltage, the mean square error of the single-cell voltage, the rate of change of the single-cell voltage, and the minimum amplitude of the single-cell voltage are calculated in real time, and the corresponding current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux, hydrogen side pressure, and oxygen side pressure are used as the control target values ​​of the fuel cell. The control inputs of the corresponding components are controlled to guide the corresponding actuators, such as: DC-DC, electronic thermostat, tail exhaust solenoid valve, hydrogen reflux pump, hydrogen proportional valve, and oxygen proportional valve, to achieve the above optimized values. Actual control can also be combined with empirical control to effectively eliminate the influence of system disturbances.

[0074] In some optional embodiments, referring to Figures 1, 2 and 3, Figure 3 is another flow chart of the fuel cell intelligent health management control method provided by the present invention, the nonlinear mapping model includes an input layer, an intermediate layer and an output layer connected in sequence, the intermediate layer includes multiple neurons; the intermediate layer is connected to a hidden layer.

[0075] Training of nonlinear mapping models, including:

[0076] Provide datasets;

[0077] Normalizing the data set to obtain a processed data set;

[0078] The data set is input into the nonlinear mapping model, and after a preset number of iterations, a trained nonlinear mapping model is obtained.

[0079] The weight update learning rate of the input layer ranges from 0.2 to 0.3, the weight update learning rate of the output layer ranges from 0.2 to 0.4, and the weight update learning rate of the hidden layer ranges from 0.1 to 0.2.

[0080] Understandably, input parameter design and output parameter model establishment prioritize accuracy, as input parameters have a significant impact on battery performance. Input samples are normalized, and weights are initially randomized. The middle layer uses 15 neurons and consists of two parts: one calculated by weighting the input layer, and the other calculated by weighting the hidden layer. The output of the middle layer is calculated using the activation function sigmoid() and enters the output layer. The output layer is designed to take into account the total stack voltage, the mean square error of the stack voltage per cell, the rate of change of the stack voltage per cell, and the minimum amplitude of the stack voltage per cell. Finally, the weights are updated, and the partial derivatives of the weights of each layer are calculated for optimal calculation, resulting in iterative convergence.

[0081] Specifically, during training, the nonlinear mapping model inputs an array of parameters. Each sample has 7 input variables, 1000 samples, and 4 output variables. The input parameters are weighted to generate intermediate layer neurons, which number 15. These neurons are then weighted to generate recurrent layer neurons to store historical information. The recurrent and intermediate layers are then combined through an activation function to form the output layer. A reverse partial derivative calculation is then performed to update the weights of each layer. Training is repeated 5000 times or until the deviation is less than or equal to 1%. The learning rate for updating the input layer weights ranges from 0.2 to 0.3, the output layer weights range from 0.2 to 0.4, and the hidden layer weights range from 0.1 to 0.2, with a bias added. The sample input variables in the big data are specifically the operating data of the fuel cell system, specifically: the fuel cell current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux, hydrogen side pressure, and oxygen side pressure. These parameters will affect the change in the water content of the fuel cell membrane electrode to a certain extent. After modeling through a nonlinear mapping model and optimization calculation, the optimal input parameters are selected as the control target value of the control system. Modeling calculation and verification can be done in either offline or online mode. The data is selected using actual operating system data, and data is read every 10 seconds. When selecting data, various representative data are considered, including power increase data, power decrease data, temperature increase data, temperature decrease data, high utilization conditions, low utilization conditions, and low, medium, and high pressure conditions.

[0082] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0083] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fuel cell intelligent health management control method, characterized in that: include: Inputting the current, temperature, hydrogen displacement, oxygen displacement, hydrogen reflux, hydrogen side pressure and oxygen side pressure of the fuel cell into the trained nonlinear mapping model, and the trained nonlinear mapping model outputs the calculation result; The calculation result is input into the objective function to find the calculated value of the total voltage of the stack, the calculated value of the mean square error of the voltage of a single cell of the stack, the calculated value of the rate of change of the voltage of a single cell of the stack, and the calculated value of the minimum amplitude of the voltage of a single cell of the stack corresponding to the maximum value of the objective function; Measure the actual value of the total voltage of the battery stack, the actual value of the mean square error of the voltage of a single cell of the battery stack, the actual value of the voltage change rate of a single cell of the battery stack, and the actual value of the minimum amplitude of the voltage of a single cell of the battery stack; the difference between the calculated value of the total voltage of the battery stack and the actual value of the total voltage of the battery stack is a first difference, the difference between the calculated value of the mean square error of the voltage of a single cell of the battery stack and the actual value of the mean square error of the voltage of a single cell of the battery stack is a second difference, the difference between the calculated value of the voltage change rate of a single cell of the battery stack and the actual value of the voltage change rate of a single cell of the battery stack is a third difference, and the difference between the calculated value of the minimum amplitude of the voltage of a single cell of the battery stack and the actual value of the minimum amplitude of the voltage of a single cell of the battery stack is a fourth difference; respectively determining whether the first difference is greater than a first threshold, whether the second difference is greater than a second threshold, whether the third difference is greater than a third threshold, and whether the fourth difference is greater than a fourth threshold; If the first difference is less than or equal to the first threshold, the second difference is less than or equal to the second threshold, the third difference is less than or equal to the third threshold, and the fourth difference is less than or equal to the fourth threshold, the current, the temperature, the hydrogen displacement, the oxygen displacement, the hydrogen reflux amount, the hydrogen side pressure and the oxygen side pressure corresponding to the calculated value of the total stack voltage, the calculated value of the mean square error of the single-cell voltage of the stack, the calculated value of the rate of change of the single-cell voltage of the stack and the calculated value of the minimum amplitude of the single-cell voltage of the stack are used as the control target values ​​of the fuel cell.

2. The intelligent health management control method for fuel cells according to claim 1, characterized in that: The objective function is calculated as follows: Y=op1×A+(1 / op2)×B+(1 / op3)×C+(1 / op4)×D; Wherein, op1 is the calculated value of the total voltage of the battery stack, A is the first weight, op2 is the calculated value of the mean square error of the voltage of a single cell of the battery stack, B is the second weight, and op3 is the calculated value of the voltage change rate of a single cell of the battery stack. C is the third weight, op4 is the calculated value of the minimum voltage amplitude of a single cell in the battery stack, and D is the fourth weight.

3. The intelligent health management control method for a fuel cell according to claim 2, characterized in that: If the first difference is greater than the first threshold, adjusting the first weight to a first optimized weight according to a fuzzy control algorithm; If the second difference is greater than the second threshold, adjusting the second weight to a second optimized weight according to a fuzzy control algorithm; If the third difference is greater than the third threshold, adjusting the third weight to a third optimized weight according to a fuzzy control algorithm; If the fourth difference is greater than the fourth threshold, the fourth weight is adjusted to a fourth optimized weight according to a fuzzy control algorithm.

4. The intelligent health management control method for a fuel cell according to claim 3, characterized in that: When at least one of the first weight, the second weight, the third weight and the fourth weight of the objective function changes to form an optimized objective function; Find the optimal value of the total voltage of the battery stack, the optimal value of the mean square error of the voltage of a single battery stack cell, the optimal value of the rate of change of the voltage of a single battery stack cell, and the optimal value of the minimum amplitude of the voltage of a single battery stack cell, which correspond to the maximum value of the optimization objective function; The current, the temperature, the hydrogen displacement, the oxygen displacement, the hydrogen reflux amount, the hydrogen side pressure and the oxygen side pressure corresponding to the optimized value of the total stack voltage, the optimized value of the mean square deviation of the single-cell voltage of the stack, the optimized value of the voltage change rate of the single-cell voltage of the stack and the optimized value of the minimum amplitude of the single-cell voltage of the stack are used as the control target values ​​of the fuel cell.

5. The intelligent health management control method for a fuel cell according to claim 1, characterized in that: The nonlinear mapping model comprises an input layer, an intermediate layer and an output layer connected in sequence, wherein the intermediate layer comprises a plurality of neurons; and the intermediate layer is connected with a hidden layer.

6. The intelligent health management control method for a fuel cell according to claim 5, characterized in that: Training the nonlinear mapping model includes: Provide datasets; Normalizing the data set to obtain a processed data set; The data set is input into the nonlinear mapping model, and after a preset number of iterations, the trained nonlinear mapping model is obtained.

7. The intelligent health management control method for a fuel cell according to claim 5 or 6, characterized in that: The weight update learning rate of the input layer ranges from 0.2 to 0.3, the weight update learning rate of the output layer ranges from 0.2 to 0.4, and the weight update learning rate of the hidden layer ranges from 0.1 to 0.2.

Images