Aging prediction method and equipment for proton exchange membrane fuel cell, and medium

Through the honey badger algorithm training of the gated cycle unit neural network model, the problem of insufficient aging prediction accuracy of proton exchange membrane fuel cells in the existing technology is solved, high-precision and low-complexity aging trend prediction is achieved, and fuel cells are extended. Lifespan and reduce maintenance costs.

CN118897965APending Publication Date: 2024-11-05LANZHOU JIAOTONG UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410941599.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-15
Publication Date
2024-11-05

AI Technical Summary

Technical Problem

The existing proton exchange membrane fuel cell aging prediction methods have insufficient accuracy and complexity, especially the prediction accuracy of data-driven methods needs to be improved, and the complexity of model construction and calculation is high.

Method used

The honey badger algorithm is used to train the gated cyclic unit neural network model, and the trained model is used to predict the aging trend of proton exchange membrane fuel cell.

Benefits of technology

Improve the aging prediction accuracy of proton exchange membrane fuel cells, establish a model with robustness and generalization capabilities, significantly improve the aging prediction accuracy of rail vehicle hybrid traction power proton exchange membrane fuel cells, extend the service life and reduce the Maintenance cost.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118897965A_ABST
    Figure CN118897965A_ABST
Patent Text Reader

Abstract

The invention discloses an aging prediction method and device of a proton exchange membrane fuel cell and a medium, and relates to the field of fuel cells, and the aging prediction method comprises the following steps: obtaining stack voltages of the proton exchange membrane fuel cell at different historical sampling time points, and obtaining a stack voltage sequence arranged according to a time historical sampling time period sequence; intercepting a training sample from the stack voltage sequence, and constructing a training set; training the gated circulation unit neural network model by using the training set and adopting a badger algorithm to obtain a trained gated circulation unit neural network model; and predicting N-n stack voltages after the current moment based on n-1 stack voltages at the current moment and before the current moment so as to predict the aging trend of the battery. According to the method, the neural network model of the gated circulation unit is trained by using the training set and adopting the badger algorithm, and the pile voltage is predicted based on the trained model, so that the aging prediction precision of the proton exchange membrane fuel cell is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of fuel cells, and in particular to an aging prediction method, device and medium for proton exchange membrane fuel cells. Background Art

[0002] With the vigorous promotion of the concept of green and environmentally friendly travel, fuel cell hybrid traction power rail vehicles have attracted widespread attention in China for their advantages such as comfortable ride and no pollution. Proton exchange membrane fuel cells are used as the main power source, which has the advantages of zero pollution, high power density and light weight, but due to their own structure, material properties and operating environment, they will cause irreversible performance degradation. Therefore, accurately predicting the aging state of proton exchange membrane fuel cells is of great significance to reducing maintenance costs, preventing catastrophic failures and extending the service life of the system.

[0003] Currently, the commonly used aging prediction methods for proton exchange membrane fuel cells can be divided into three types:

[0004] 1. Model-driven method, which mainly uses a specific mechanism model or semi-empirical model to predict the stack voltage of the fuel cell. Currently, this type of method is mainly based on filtering algorithms and semi-empirical models;

[0005] 2. Data-driven methods, which mainly build aging models by learning a large amount of experimental data, are model-free methods. Currently, this type of method is mainly based on sparse kernel machines and artificial neural network methods;

[0006] 3. Hybrid method, which refers to the combination of model-driven method and data-driven method, or the integration of multiple data-driven methods, in the hope of further improving the accuracy and reliability of predictions.

[0007] Among the above three aging prediction methods for proton exchange membrane fuel cells, the model-driven method can obtain aging parameters to characterize the internal aging state, but the aging process involves the combined effects of multiple factors, making it difficult to establish an accurate aging model; the hybrid method can achieve high prediction accuracy while obtaining fuel cell aging parameters, but the complexity of model building and calculation has also increased with the improvement of prediction accuracy. Compared with the model-driven method and the hybrid method, the data-driven method can flexibly represent the aging characteristics observed in the aging voltage data, and does not involve the aging mechanism inside the fuel cell, but the prediction accuracy needs to be improved. Summary of the invention

[0008] The purpose of the present application is to provide a proton exchange membrane fuel cell aging prediction method, device and medium, which can achieve accurate prediction of the stack voltage of the proton exchange membrane fuel cell.

[0009] To achieve the above objectives, this application provides the following solutions:

[0010] In a first aspect, the present application provides an aging prediction method for a proton exchange membrane fuel cell, the aging prediction method for a proton exchange membrane fuel cell comprising: obtaining the stack voltage of the proton exchange membrane fuel cell at different historical sampling time points, and obtaining a stack voltage sequence arranged in order according to the time historical sampling time period; intercepting training samples in the stack voltage sequence to construct a training set; the first n stack voltages in the training sample are input data, and the last Nn stack voltages in the training sample are output data, and N is the number of stack voltages contained in the training sample; using the training set to train a gated recurrent unit neural network model using a honey badger algorithm to obtain a trained gated recurrent unit neural network model; using the trained gated recurrent unit neural network model, based on the n-1 stack voltages at the current moment and before the current moment, predicting the Nn stack voltages after the current moment; and predicting the aging trend of the proton exchange membrane fuel cell based on the predicted Nn stack voltages after the current moment.

[0011] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aging prediction method for a proton exchange membrane fuel cell described in any one of the above.

[0012] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned methods for predicting aging of a proton exchange membrane fuel cell.

[0013] According to the specific embodiments provided in the present application, the present application discloses the following technical effects: The present application provides a method, device and medium for predicting the aging of a proton exchange membrane fuel cell, by using a training set to adopt a honey badger algorithm to train a gated recurrent unit neural network model, and predicting the stack voltage based on the trained model, thereby improving the aging prediction accuracy of the proton exchange membrane fuel cell. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 A schematic flow chart of an aging prediction method for a proton exchange membrane fuel cell provided in an embodiment of the present application.

[0016] Figure 2 A topological structure of a hybrid traction power system for a rail vehicle equipped with a proton exchange membrane fuel cell is provided in an embodiment of the present application.

[0017] Figure 3 This is a graph of the stack voltage sequence after denoising under static conditions provided in an embodiment of the present application.

[0018] Figure 4 This is a graph of the stack voltage sequence after denoising under dynamic conditions provided in an embodiment of the present application.

[0019] Figure 5 The stack voltage prediction result provided in the embodiment of the present application is under static conditions with 700h as the prediction starting point.

[0020] Figure 6 The stack voltage prediction result provided in the embodiment of the present application under dynamic conditions with 700h as the prediction starting point.

[0021] Figure 7 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0024] like Figure 1 As shown, this embodiment provides an aging prediction method for a proton exchange membrane fuel cell, which specifically includes:

[0025] S1. Obtain the stack voltage of the proton exchange membrane fuel cell at different historical sampling time points, and obtain a stack voltage sequence arranged in the order of the historical sampling time periods. The proton exchange membrane fuel cell is a proton exchange membrane fuel cell of a hybrid traction power system of a rail vehicle.

[0026] Furthermore, the stack voltage of the proton exchange membrane fuel cell at different historical sampling time points is collected under any one of a static operating condition or a dynamic operating condition.

[0027] In actual application, Figure 2The topology of a hybrid traction power system for a rail vehicle equipped with a proton exchange membrane fuel cell is shown. For the vehicle hybrid traction power system, the proton exchange membrane fuel cell is used as the main power source to provide the power required by the load, and the lithium battery and supercapacitor are used as auxiliary power sources to ensure the energy density and power density requirements of the tram.

[0028] By preprocessing the aging data set of proton exchange membrane fuel cells in the hybrid traction power system of rail vehicles, a static operating condition data set DS1 or a dynamic operating condition data set DS2 after preprocessing (denoising) is obtained; the static operating condition data set DS1 or the dynamic operating condition data set DS2 in the proton exchange membrane fuel cell in the hybrid traction power system of rail vehicles is obtained, and the characteristic parameters of the data set DS1 or DS2 include the single cell voltage, the stack voltage, the current, and the temperature and pressure of the reaction gas inlet and outlet.

[0029] S2. Extract training samples from the stack voltage sequence to construct a training set; the first n stack voltages in the training samples are input data, the last Nn stack voltages in the training samples are output data, and N is the number of stack voltages included in the training samples.

[0030] Optionally, before step S2, the method further includes: denoising the battery stack voltage sequence to obtain a denoised battery stack voltage sequence.

[0031] Furthermore, the specific process of the denoising process is as follows:

[0032] The stack voltage sequence is reconstructed at preset time intervals to obtain a reconstructed stack voltage sequence.

[0033] The Gaussian moving average filtering method is used to smooth the reconstructed stack voltage sequence to obtain the denoised stack voltage sequence.

[0034] In actual application, Figure 3-Figure 4 As shown, in order to solve the problems of excessive calculation, long training time and many noises and spikes in the data, the data is preprocessed (denoising), and then the preprocessed DS1 and DS2 data are divided into training set and validation set with 700h as the prediction starting point.

[0035] S3. Using the training set, the gated recurrent unit neural network model is trained using the Honey Badger algorithm to obtain a trained gated recurrent unit neural network model.

[0036] Step S3 specifically includes:

[0037] Step S31. Using the network parameters of the gated recurrent unit neural network as population individuals, the population of the honey badger algorithm is initialized to obtain initialized population individuals; the network parameters include at least: the number of hidden units and the L2 regularization parameter.

[0038] Optionally, the network parameters also include: training cycle and learning rate. The training cycle, number of hidden units, L2 regularization parameter and learning rate respectively determine the convergence degree, expression ability, generalization ability and stability of the gated recurrent unit neural network model.

[0039] In actual application, assuming that the population size is N and the problem dimension is D, the initialization formula is as follows:

[0040] x ij =lb j +r 1 .(ub j -lb j ).

[0041] Where, i = 1, 2, ..., N; j = 1, 2, ..., D. ij represents the j-th dimension variable of individual i, ub j and lb j Respectively represent the upper and lower bounds of the j-th dimension variable, r 1 is a random number between [0,1], f i The smaller the value, the better the individuals in the population.

[0042] Step S32: Input the training set into the gated recurrent unit neural network model to calculate the predicted value of the stack voltage of the training samples in the training set.

[0043] Step S33. Taking the predicted values ​​of the battery stack voltages of all training samples and the mean square error of the battery stack voltages of the corresponding training samples in the training set as the fitness function of the initialized population individuals, the fitness of each initialized population individual is calculated.

[0044] Optionally, the calculation formula of the fitness function is as follows:

[0045]

[0046] Among them, f i is the fitness function of the i-th individual in the population, is the stack voltage predicted by the gated recurrent unit neural network optimized by the i-th population individual for the j-th training sample in the training set, y j represents the output data of the jth training sample in the training set, and m is the number of training samples in the training set.

[0047] Step S34. Determine the local optimal solution of the qth iteration according to the fitness of the initialized population individuals of the qth iteration; the local optimal solution is the population individual with the lowest fitness in this iteration process.

[0048] Step S35. When the local optimal solution of the qth iteration is less than the global optimal solution, the global optimal solution is updated; the global optimal solution is the population individual with the lowest fitness in all the iterations.

[0049] Step S36. Based on the population individuals of the qth iteration, calculate the prediction result of the qth iteration, update the population individuals, and construct the gated recurrent unit neural network optimized by different population individuals for the next iteration.

[0050] Step S37. Based on the gated recurrent unit neural network optimized by individuals from different populations in the next iteration, calculate the fitness of the new population individuals until the maximum number of iterations is reached and output the global optimal solution.

[0051] Step S38. Based on the global optimal solution, construct a trained gated recurrent unit neural network model.

[0052] In practical applications, the root mean square error is used as the fitness function to evaluate the quality of individuals in the population and calculate the fitness function value of each individual in the population.

[0053] For each individual x i , construct a gated recurrent unit using the network parameters it represents, and train it on the training set. Then use the trained gated recurrent unit to predict the test set, and calculate the root mean square error between the predicted value and the true value, which is used as the fitness function value f of the individual in the population i Among them, multiple groups of historical characteristic data are used as input data, and the stack voltage is used as output data.

[0054] The global optimal solution for the new round of iteration is determined by evaluating the fitness function value of each individual in the population.

[0055] The population individual with the smallest fitness function value is selected as the local optimal solution of this iteration, and compared with the historical local optimal solution. If the optimal solution of this iteration is better, it will be updated to the global optimal solution.

[0056] By selecting the update strategy for the individual positions of the population, the fitness function value of the new round of global optimal solution is calculated.

[0057] According to the size of r, select the update strategy of the individual position of the population and calculate the fitness function value f of the new population individual jnew , the calculation formula is as follows:

[0058]

[0059] Among them, f inew is the fitness function of the jth new population individual, is the stack voltage predicted by the gated recurrent unit neural network optimized by the jth new population individual in the i-th training sample in the training set, y i represents the output data of the i-th training sample in the training set, and m is the number of training samples in the training set.

[0060] The update strategy of the individual position of the population is divided into two stages: mining mode and honey collection mode: if r < 0.5 (r is a random number between [0, 1]), the individual position of the population is updated according to the mining mode to obtain x inew ; If r>0.5 (r is a random number between [0,1]), the individual positions of the population are updated according to the honey collection mode to obtain x inew The two-stage mathematical model is as follows.

[0061] x inew =x prey +F.β.I i .x prey +Fr 3 .α.d i .|cos(2πr 4 ).[1-cos(2πr 5 )]|.

[0062] Among them, the odor intensity I i The calculation formula for calculating the density factor α is as follows.

[0063]

[0064] Where S is the concentrated intensity, d i is the distance between the prey and the i-th honey badger, r 2 is a random number between [0,1], x prey Represents the best individual in the population.

[0065]

[0066] Among them, T is the maximum number of iterations, t is the current number of iterations, and C is a constant not less than 1, usually set to 2.

[0067] Among them, x inew is the updated position of individual i in the population, β≥1, represents the predation ability of the honey badger, usually set to 6, r 3 , r 4 , r 5They are all different random numbers between [0,1]. F is a flag used to change the search direction. The calculation formula is as follows.

[0068]

[0069] Among them, r 6 is a random number between [0,1].

[0070] x inew =x prey +Fr 7 .α.d i .

[0071] Among them, r 7 It is also a random number between [0,1].

[0072] When the maximum number of iterations is reached, the global optimal solution is output, and a gated recurrent unit neural network is constructed through the network parameters represented by the optimal solution to predict the long-term aging trend (stack voltage) of the proton exchange membrane fuel cell.

[0073] Among them, the comparison rule for determining the optimal population individual is: if f inew <f i , then x i =x inew , f i =f inew If f inew <f prey , then x prey =x inew , f prey =f inew .

[0074] S4. Using the trained gated recurrent unit neural network model, based on the voltages of the n-1 stacks at the current moment and before the current moment, predict the voltages of the Nn stacks after the current moment; wherein Figure 5-Figure 6 The predicted results of the stack voltage under static conditions and under static conditions are shown.

[0075] S5. Predict the aging trend of the proton exchange membrane fuel cell based on the predicted Nn stack voltages after the current moment.

[0076] The technical effects of the present application are as follows: The present application uses the Honey Badger algorithm to train the gated recurrent unit neural network model using the training set, and predicts the stack voltage based on the trained model, thereby improving the prediction accuracy of the stack voltage of the proton exchange membrane fuel cell, so that the established aging prediction model has strong robustness and generalization ability, significantly improving the aging prediction accuracy of the hybrid traction power proton exchange membrane fuel cell of the rail vehicle, thereby improving the durability of the proton exchange membrane fuel cell, reducing maintenance costs, and extending the service life. The specific effects of the present application are as follows: 1. The present application establishes a prediction model based on the gated recurrent unit neural network, and uses its advantages in processing long-term and short-term dependencies to improve the prediction accuracy. In order to further optimize the hyperparameter group of the gated recurrent unit and improve the generalization ability and stability of the model, the Honey Badger optimization algorithm is introduced as a hyperparameter optimization method to achieve hyperparameter adaptive adjustment. 2. The present application combines the advantages of the Honey Badger optimization algorithm and the gated recurrent unit neural network, which can effectively deal with the unsteady, multivariable and nonlinear characteristics of the proton exchange membrane fuel cell, and realizes the long-term aging trend prediction of the hybrid traction power proton exchange membrane fuel cell of the rail vehicle with both accuracy and reliability.

[0077] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video tag processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting the aging of a proton exchange membrane fuel cell is implemented.

[0078] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0079] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0080] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0081] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0082] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0083] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for predicting aging of a proton exchange membrane fuel cell, characterized in that: The aging prediction method of the proton exchange membrane fuel cell comprises: Obtaining the stack voltage of the proton exchange membrane fuel cell at different historical sampling time points, and obtaining a stack voltage sequence arranged in the order of the historical sampling time periods; A training sample is intercepted from the stack voltage sequence to construct a training set; the first n stack voltages in the training sample are input data, the last Nn stack voltages in the training sample are output data, and N is the number of stack voltages included in the training sample; Using the training set and the Honey Badger algorithm to train the gated recurrent unit neural network model, to obtain a trained gated recurrent unit neural network model; Using the trained gated recurrent unit neural network model, based on the voltages of the n-1 battery stacks at and before the current moment, the voltages of the Nn battery stacks after the current moment are predicted; The aging trend of the proton exchange membrane fuel cell is predicted based on the predicted Nn stack voltages after the current moment.

2. The aging prediction method for proton exchange membrane fuel cells according to claim 1, characterized in that: The gated recurrent unit neural network model is trained using the training set and the honey badger algorithm to obtain a trained gated recurrent unit neural network model, specifically including: Using the network parameters of the gated recurrent unit neural network as population individuals, the population of the honey badger algorithm is initialized to obtain the initialized population individuals; the network parameters at least include: the number of hidden units and the L2 regularization parameter; The training set is input into the gated recurrent unit neural network model to calculate the predicted value of the stack voltage of the training samples in the training set; The fitness function of each initialized individual in the population is taken as the fitness function of the predicted value of the stack voltage of all training samples and the mean square error of the stack voltage of the corresponding training samples in the training set, and the fitness of each initialized individual in the population is calculated; According to the fitness of the population individuals after the initialization of the qth iteration, determine the local optimal solution of the qth iteration; the local optimal solution is the population individual with the lowest fitness in this iteration process; When the local optimal solution of the qth iteration is less than the global optimal solution, the global optimal solution is updated; the global optimal solution is the population individual with the lowest fitness in all previous iterations; Based on the population individuals of the qth iteration, the prediction results of the qth iteration are calculated, the population individuals are updated, and the gated recurrent unit neural network optimized by different population individuals of the next iteration is constructed; Based on the gated recurrent unit neural network optimized by individuals from different populations in the next iteration, the fitness of the new population individuals is calculated until the maximum number of iterations is reached and the global optimal solution is output; Based on the global optimal solution, a trained gated recurrent unit neural network model is constructed.

3. The aging prediction method for proton exchange membrane fuel cells according to claim 2, characterized in that: The calculation formula of the fitness function is as follows: Among them, f i is the fitness function of the i-th individual in the population, is the stack voltage predicted by the gated recurrent unit neural network optimized by the i-th population individual for the j-th training sample in the training set, y j represents the output data of the jth training sample in the training set, and m is the number of training samples in the training set.

4. The aging prediction method of a proton exchange membrane fuel cell according to claim 1, characterized in that: The stack voltage of the proton exchange membrane fuel cell at different historical sampling time points is collected under any one of a static operating condition or a dynamic operating condition.

5. The aging prediction method of proton exchange membrane fuel cell according to claim 1, characterized in that: The training samples are intercepted from the stack voltage sequence, and before constructing the training set, the following are also included: The stack voltage sequence is denoised to obtain a denoised stack voltage sequence.

6. The aging prediction method for proton exchange membrane fuel cells according to claim 5, characterized in that: The specific process of the denoising process is as follows: Reconstructing the data of the battery stack voltage sequence at a preset time interval to obtain a reconstructed battery stack voltage sequence; The Gaussian moving average filtering method is used to smooth the reconstructed stack voltage sequence to obtain the denoised stack voltage sequence.

7. The aging prediction method for proton exchange membrane fuel cells according to claim 6, characterized in that: The time interval is 1 hour.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the aging prediction method for a proton exchange membrane fuel cell according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the aging prediction method for a proton exchange membrane fuel cell according to any one of claims 1 to 7 are implemented.