Proton exchange membrane fuel cell voltage decline prediction method, equipment, medium and product
By preprocessing and empirical modal decomposition of the historical voltage data of the proton exchange membrane fuel cell, combined with the grayscale prediction model, the problem of poor voltage decay prediction accuracy in the prior art is solved, and high-precision prediction of the voltage decay of the fuel cell is achieved.
Patent Information
- Application Number
- CN202510228561.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-28
- Publication Date
- 2025-05-30
AI Technical Summary
The prediction accuracy of the proton exchange membrane fuel cell voltage decay in the prior art is poor, and the impact of different frequency operating conditions on voltage decay is not effectively considered.
By preprocessing the historical voltage data output by the proton exchange membrane fuel cell, multiple eigenmodal functions and trend terms are generated using the empirical modal decomposition (EMD) algorithm, frequency classification and reconstruction are performed, and the fuel cell voltage decay prediction results are determined in combination with the grayscale prediction model.
It realizes high-precision prediction of the voltage decay of the proton exchange membrane fuel cell, and can more carefully capture the influencing factors under different frequency conditions, improving the accuracy of the prediction results.
Smart Images
Figure CN120064981A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of fuel cells, and particularly to a method, device, medium and product for predicting voltage degradation of a proton exchange membrane fuel cell. Background Art
[0002] Proton Exchange Membrane Fuel Cells (PEMFCs) are widely regarded as an effective solution to increasing energy demand and environmental pollution due to their characteristics such as zero carbon emissions, ecological friendliness, and high efficiency, and are applied in multiple fields such as transportation and power supply. However, their performance is vulnerable to material degradation and adverse operating conditions, resulting in durability and reliability issues, which limit their widespread application. To improve energy conversion efficiency and reliability, it is crucial to extend their service life and enhance stability. By predicting their degradation data, the aging trend can be predicted more accurately and the remaining service life can be estimated.
[0003] Although traditional grey prediction models have been used to predict the voltage degradation of proton exchange membrane fuel cells, these grey prediction models consider different frequency operating conditions (i.e., fluctuations and changes in low frequency and high frequency) together, and only perform degradation prediction based on the original voltage data of the proton exchange membrane fuel cell, without considering the influence of the factor causing different frequency operating conditions on the voltage degradation of the proton exchange membrane fuel cell, thus resulting in poor prediction accuracy of the voltage degradation of the proton exchange membrane fuel cell. Summary of the Invention
[0004] The purpose of the present application is to provide a method, device, medium and product for predicting voltage degradation of a proton exchange membrane fuel cell, which can solve the problem of poor prediction accuracy of the voltage degradation of the proton exchange membrane fuel cell in the prior art.
[0005] To achieve the above purpose, the present application provides the following solutions:
[0006] In a first aspect, the present application provides a method for predicting voltage degradation of a proton exchange membrane fuel cell, including:
[0007] Preprocessing the historical voltage data output by the proton exchange membrane fuel cell to obtain preprocessed data;
[0008] Performing empirical mode decomposition on the preprocessed data to generate a plurality of intrinsic mode functions and a trend term;
[0009] Classifying the frequencies of the plurality of intrinsic mode functions to obtain a high-frequency sequence and a low-frequency sequence;
[0010] Reconstructing the high-frequency sequence and the low-frequency sequence respectively to obtain a total high-frequency mode and a total low-frequency mode;
[0011] Determine the predicted result of fuel cell voltage degradation based on the total high-frequency mode, the total low-frequency mode, the trend term, and the gray prediction model.
[0012] In a second aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to implement the method for predicting the voltage degradation of a proton exchange membrane fuel cell described above.
[0013] In a third aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the method for predicting the voltage degradation of a proton exchange membrane fuel cell described above.
[0014] In a fourth aspect, the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the method for predicting the voltage degradation of a proton exchange membrane fuel cell described in any one of the above.
[0015] According to the specific embodiments provided by the present application, the following technical effects are disclosed in the present application:
[0016] The present application provides a method, device, medium, and product for predicting the voltage degradation of a proton exchange membrane fuel cell. First, preprocess the historical voltage data output by the proton exchange membrane fuel cell to obtain preprocessed data. Then, perform empirical mode decomposition on the preprocessed data to generate a plurality of intrinsic mode functions and a trend term. Preprocessing the historical voltage data and combining empirical mode decomposition can decompose the historical voltage data into empirical modes corresponding to different frequencies, avoiding the use of only the original voltage data in traditional methods and improving the accuracy of the subsequent predicted results of fuel cell voltage degradation. Further, classify the frequencies of the plurality of intrinsic mode functions to obtain a high-frequency sequence and a low-frequency sequence, and respectively reconstruct the high-frequency sequence and the low-frequency sequence to obtain a total high-frequency mode and a total low-frequency mode, realizing the distinction between low-frequency operating conditions and high-frequency operating conditions. Finally, determine the predicted result of fuel cell voltage degradation based on the total high-frequency mode, the total low-frequency mode, the trend term, and the gray prediction model, which can more carefully capture the influencing factors under different frequency operating conditions, and ultimately achieve high-precision prediction of the voltage degradation of proton exchange membrane fuel cells. Description of the Drawings
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.
[0018] Figure 1 It is a schematic flowchart of a method for predicting the voltage degradation of a proton exchange membrane fuel cell provided in an embodiment of the present application;
[0019] Figure 2 It is a schematic diagram of empirical mode decomposition of a method for predicting the voltage degradation of a proton exchange membrane fuel cell provided in an embodiment of the present application. Detailed implementation manners
[0020] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.
[0021] To make the above objects, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below in conjunction with the drawings and specific implementation manners.
[0022] As Figure 1 shown, the present application provides a method for predicting the voltage degradation of a proton exchange membrane fuel cell, including:
[0023] Step 101: Preprocess the historical voltage data output by the proton exchange membrane fuel cell to obtain preprocessed data.
[0024] In some embodiments, step 101 specifically includes: within a preset time period, historical voltage data output by the proton exchange membrane fuel cell is acquired at preset time intervals; the preset time period is a known interval, and the period after the preset time period is an extrapolation interval; the historical voltage data acquired within the known interval is used to construct the gray prediction model; the historical voltage data acquired within the extrapolation interval is used for extrapolation prediction of the gray prediction model; according to the historical voltage data output by the proton exchange membrane fuel cell each time, the historical voltage data is determined; the historical voltage data is denoised and outliers are removed to obtain the preprocessed data.
[0025] Among them, historical voltage data can be taken once every 0.5h.
[0026] Step 102: Perform empirical mode decomposition on the preprocessed data to generate a plurality of intrinsic mode functions and a trend term.
[0027] In practical applications, the empirical mode decomposition (EMD) algorithm is used to decompose the preprocessed data into decomposition data.
[0028] Among them, the conditions for the EMD algorithm include: the signal has at least two extreme points, one maximum and one minimum. That is, in the iterative process of EMD, the upper and lower envelopes need to be generated by spline interpolation, and these envelopes are constructed based on the extreme points of the signal. If there are not enough extreme points in the signal, the EMD algorithm will not be able to correctly extract the Intrinsic Mode Function (IMF).
[0029] The characteristic time scale is defined by the time between two extreme points. That is, each IMF in the EMD algorithm represents an oscillation mode in the signal, and the characteristic time scale of the oscillation mode can be defined by the distance between adjacent extreme points. Specifically, the time interval between two adjacent extreme points reflects the period or frequency of this IMF.
[0030] If the data lacks extreme points but has deformation points, the extreme points can be obtained by differentiating the data once or several times, and then the decomposition result can be obtained by integration. That is to say, in some cases, the signal may lack obvious extreme points, especially when the signal is relatively smooth or has a slow-changing trend. In this case, directly applying the EMD algorithm may encounter difficulties because effective envelope lines cannot be generated. Therefore, the extreme points can be obtained by differentiating the data once or several times, and then the decomposition result can be obtained by integration. Among them, the signal is the preprocessed voltage data.
[0031] Specifically, the decomposition process using the EMD algorithm is as follows:
[0032] Step 1: Assume that the preprocessed data is x(t). Use spline interpolation to fit the maximum and minimum points of the preprocessed data to form the upper and lower envelope lines, and calculate the average envelope line x 0 (t). Then subtract the mean from the preprocessed data to obtain a new signal h 1 (t). The expression of h 1 (t) is as follows.
[0033] h 1 (t) = x(t) - x 0 (t).
[0034] Step 2: Judge whether the new signal h 1 (t) can meet the conditions of IMF. If it meets, then h 1 (t) is the first IMF component. If it does not meet, then continue to loop the first step until h 1 (t) meets the characteristic conditions of IMF.
[0035] Step 3: Separate the IMF in the preprocessed data 1 to obtain r 1(t), the new signal expression is as follows, and loop until the second IMF component is obtained.
[0036] r 1 (t) = x(t) - IMF 1 。
[0037] Step 4: Repeat the above steps until the nth IMF component is obtained. Finally, the result of empirical mode decomposition can be expressed as x(t) = IMF n +r n 。
[0038] where r n is the trend term obtained after separating the IMF n times, representing the central trend of the signal; n = 1, 2, 3...
[0039] Step 103: Classify the multiple intrinsic mode functions by frequency to obtain a high-frequency sequence and a low-frequency sequence.
[0040] In some embodiments, Step 103 specifically includes Steps 201 - 204:
[0041] Step 201: Starting from the first generated intrinsic mode function in the order of generation of the multiple intrinsic mode functions, obtain i intrinsic mode functions.
[0042] Step 202: Based on the i intrinsic mode functions, construct i indicators; where the i-th indicator is the sum of the i intrinsic mode functions.
[0043] Step 203: Calculate the mean of the i indicators.
[0044] Step 204: Based on the mean and the first threshold, determine the high-frequency sequence and the low-frequency sequence.
[0045] In some embodiments, Step 204 specifically includes: judging whether the difference between the mean and 0 is greater than or equal to the first threshold to obtain a first result; if the first result is yes, determine the first i - 1 intrinsic mode functions as the high-frequency sequence and the intrinsic mode functions after the (i - 1)-th intrinsic mode function as the low-frequency sequence; if the first result is no, return to the step "Starting from the first generated intrinsic mode function in the order of generation of the multiple intrinsic mode functions, obtain i intrinsic mode functions", and update i to i + 1 until the difference between the mean and 0 is greater than or equal to the first threshold.
[0046] Among them, since the IMF components need to satisfy that the upper and lower envelopes are locally symmetric with respect to the time axis. For high-frequency IMF components, the upper and lower envelopes are basically obtained by connecting numerous signal peak points. Therefore, the symmetry of the envelopes means that the IMF component data is basically symmetric and the data mean approaches 0. For low-frequency IMF components, the signal period is large, and the envelopes are obtained by interpolating a small number of peaks. The envelope trend deviates greatly from the trend of the original signal. Therefore, when the envelopes are symmetric, the signal components are often not symmetric or even deviate far. In this case, it is natural that the mean of the IMF components is difficult to ensure to be 0.
[0047] Therefore, denote IMF1 as index 1, IMF1 + IMF2 as index 2, and so on. Add the first i IMFs to obtain index i. Calculate the mean of indices 1 to i, and perform a τ-test on whether this mean is significantly different from 0. When it is detected that a certain mean is significantly different from 0, determine the first i - 1 intrinsic mode functions as the high-frequency sequence, and determine the intrinsic mode functions after the (i - 1)-th intrinsic mode function as the low-frequency sequence.
[0048] Among them, the process of the τ-test is as follows.
[0049] Null hypothesis (H0): The mean is not significantly different from 0; alternative hypothesis (H1): The mean is significantly different from 0. Select the significance level α to be 0.05, that is, the preset threshold. Take m samples in index i. Calculate the statistic T:
[0050]
[0051] Among them, is the mean of the selected i intrinsic mode functions; μ is the assumed population mean, which is 0 here; s is the standard deviation.
[0052] Find the P-value according to the T value in the T-distribution table. If the P-value is less than the significance level α, reject the null hypothesis and consider that the population mean is significantly different from 0. If the P-value is greater than or equal to the significance level α, do not reject the null hypothesis and consider that the population mean is not significantly different from 0.
[0053] Step 104: Reconstruct the high-frequency sequence and the low-frequency sequence respectively to obtain the total high-frequency mode and the total low-frequency mode.
[0054] In some embodiments, step 104 specifically includes: adding each intrinsic mode function in the high-frequency sequence to obtain the total high-frequency mode; adding each intrinsic mode function in the low-frequency sequence to obtain the total low-frequency mode.
[0055] Step 105: Determine the fuel cell voltage degradation prediction result according to the total high-frequency mode, the total low-frequency mode, the trend term, and the gray prediction model.
[0056] Among them, the gray prediction model includes the GM(1,1) gray model.
[0057] In some embodiments, step 105 specifically includes steps 301-306.
[0058] Step 301: Input the total high-frequency mode, the total low-frequency mode, and the trend term as input data into the gray prediction model to obtain prediction data; the prediction data includes high-frequency prediction data, low-frequency prediction data, and trend term prediction data.
[0059] Step 302: Calculate the difference between the prediction data and the input data to obtain a residual data sequence.
[0060] Step 303: Input the absolute values of the elements in the residual data sequence into the gray prediction model to obtain a residual prediction value.
[0061] Step 304: Based on the known interval and the extrapolation interval, correct the residual prediction value to obtain a corrected residual prediction value.
[0062] In some embodiments, step 304 specifically includes: when the residual prediction value is within the known interval, correct the residual prediction value according to the positive and negative states of each element in the residual data sequence to determine the corrected residual prediction value; when the residual prediction value is within the extrapolation interval, correct the residual prediction value by the Markov residual correction method to determine the corrected residual prediction value.
[0063] Step 305: Add the corrected residual prediction value to the prediction data to obtain the final prediction data.
[0064] Step 306: Use the final prediction data as the fuel cell voltage degradation prediction result.
[0065] Specifically, according to steps 101-104, the total high-frequency mode IMFg, the total low-frequency mode IMFd, and the trend term r n Three groups of data sequences are obtained. Build gray prediction models according to the three groups of data sequences respectively. Taking the trend term as an example, it is as follows.
[0066] Let the trend term be X 0 It is expressed as: X 0 ={X 0 (1), X 0 (2), X 0 (3), …, X 0 (n)}.
[0067] According to Perform superposition to obtain the generated sequence X 1 , where X1 = {X 1 (1), X 1 (2), X 1 (3), …, X 1 (n)}. The specific process is as follows:
[0068] Take the adjacent mean of the generated sequence:
[0069]
[0070] where z 1 is the background value; X 1 (k) and X 1 (k - 1) are the values at the k-th moment and the (k - 1)-th moment in the generated sequence X 1 ; k is the index of the time series. k is used to describe the exponential change of the system state over time. In the grey prediction model, it is mainly used to simulate the growth or decay trend of data, where the adjacent mean is the background value.
[0071] Through X 0 + az 1 = b, we get u = (a, b) T = (B T B) -1 B T Y.
[0072] where a is the development coefficient in the grey prediction model; b is the grey action quantity in the grey prediction model; u is a parameter vector; B is the design matrix; B contains the background value z 1 and a constant term, and the constant term is set to 1; T is the transpose of the matrix; Y is the vector representation of the trend term X 0 . Specifically, the grey prediction model calculates a and b through the least squares method.
[0073]
[0074] where X 1 (k + 1) is the value at the (k + 1)-th moment in the generated sequence X 1 ; e is a constant.
[0075] It should be noted that in the grey prediction model, the accumulated value refers to the sequence obtained after the original data undergoes the accumulated generation operation. The predicted accumulated value refers to the accumulated value at a future time point predicted by the grey prediction model based on the existing data.
[0076] Substitute the generated sequence X 1 into the grey prediction model to calculate the estimated values of each item in the generated sequence
[0077] The estimated values of each item in the generated sequence are successively subtracted to obtain the trend item prediction data. Among them, the trend item prediction data at the (k + 1)-th moment is
[0078] At the same time, let
[0079] Calculate the residual data e′, and the formula is as follows:
[0080]
[0081] Therefore, a set of residual data sequences is obtained as:
[0082]
[0083] Among them, represents a set of residual data, that is, a set of residual data sequences.
[0084] Put After absolute value processing of the elements in, input them into the gray prediction model to obtain the residual prediction value
[0085] In the known interval, according to the positive and negative states of each element in the residual data sequence before absolute value processing, the corresponding residual prediction value can be corrected. The corrected residual prediction value is
[0086]
[0087] Among them, δ t 1 is the residual correction coefficient in the known interval; is the residual correction coefficient in the extrapolation interval.
[0088]
[0089] Among them, δ t 1 is the residual correction coefficient in the known interval, and e t ′ is the residual data in the known interval.
[0090] In the extrapolation interval, the Markov residual correction method is introduced to correct the residual prediction value. In the Markov residual correction process, the possibility that a thing changes from one state to another is called the transition probability. Since the residual value has only two states, positive and negative, at a certain moment, if the residual value is positive, the state value is recorded as 1; otherwise, the state value is recorded as 0.
[0091] The transition probability from state v to state y is denoted as P vy . P vy The calculation formula is:
[0092]
[0093] v = 0, 1;
[0094] y = 0, 1;
[0095] Where M v is the number of times the residual value is in state v, and M vy is the number of times the residual value changes from state v to state y.
[0096] Initial state of the state transition matrix State distribution vector after r-step transition Where and are the probabilities that the residual value is in state 1 and state 0 respectively before time t, that is, the probabilities that the residual value is positive and negative; P (r) is the probability matrix after r-step transition composed of P vy . The expression of P (r) is:
[0097]
[0098] Where each element in is the probability that the residual prediction value transfers from one state to another state, and is determined by .
[0099] According to the size relationship of the elements in, the Markov residual correction coefficient at the t n+r th moment can be determined That is
[0100]
[0101] Where are the probabilities that the residual value is in state 1 and state 0 respectively after r-step transition at time t; is the residual correction coefficient of the extrapolation interval.
[0102] After Markov residual correction, the trend term prediction data
[0103] Similarly, the high-frequency prediction data and the low-frequency prediction data are obtained by the above method and will not be elaborated here.
[0104] In summary, according to the final prediction data Taking the final predicted data as the prediction result of fuel cell voltage degradation, this application is based on the GM(1,1) gray model and the Markov residual correction method as the Markov gray prediction model, which can more precisely capture the influencing factors under different frequency conditions, thereby achieving high-precision prediction of proton exchange membrane fuel voltage degradation.
[0105] This application proposes a method for predicting proton exchange membrane fuel cell voltage degradation. By obtaining the output voltage data of the proton exchange membrane fuel cell and performing preprocessing; decomposing through the EMD algorithm to obtain multiple intrinsic mode functions IMF; classifying the multiple intrinsic mode functions IMF by frequency to obtain high-frequency sequences, low-frequency sequences, and a trend term, and respectively reconstructing the high-frequency sequences and low-frequency sequences to obtain the total high-frequency mode and the total low-frequency mode; based on the total high-frequency mode, total low-frequency mode, and trend term obtained above, respectively build gray prediction models to predict voltage degradation data, and sum the three predicted data to obtain the final predicted data of historical voltage data, realizing accurate prediction of fuel cell voltage degradation. That is to say, based on the EMD algorithm, a signal can be decomposed into empirical modes of different frequencies. By analyzing these decomposed modes, the influencing factors under different frequency conditions can be more precisely captured, thereby achieving high-precision prediction of proton exchange membrane fuel voltage degradation.
[0106] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the above method is implemented.
[0107] 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 above method is implemented.
[0108] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the above method is implemented.
[0109] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.
[0110] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. 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 methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random access memories (ReRAM), magnetoresistive random access memories (MRAM), ferroelectric random access memories (FRAM), phase change memories (PCM), graphene memories, etc. Volatile memories can include random access memories (RAM) or external cache memories, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.
[0111] The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.
[0112] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, 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, it should be considered as the scope described in this specification.
[0113] Specific examples are used in this article to elaborate on the principles and implementation manners of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.
Claims
1. A method for predicting voltage decay of a proton exchange membrane fuel cell, characterized in that: include: Preprocessing historical voltage data output by the proton exchange membrane fuel cell to obtain preprocessed data; Performing empirical mode decomposition on the preprocessed data to generate multiple intrinsic mode functions and a trend term; Frequency classification of the multiple intrinsic mode functions to obtain high-frequency sequences and low-frequency sequences; Reconstructing the high-frequency sequence and the low-frequency sequence respectively to obtain a total high-frequency mode and a total low-frequency mode; A fuel cell voltage decay prediction result is determined according to the total high-frequency mode, the total low-frequency mode, the trend item and the grayscale prediction model.
2. The method for predicting voltage decay of a proton exchange membrane fuel cell according to claim 1, characterized in that: The historical voltage data output by the proton exchange membrane fuel cell is preprocessed to obtain preprocessed data, specifically including: Within a preset time period, historical voltage data output by a proton exchange membrane fuel cell is obtained once at every preset time interval; the preset time period is a known interval, and the interval after the preset time period is an extrapolated interval; the historical voltage data obtained in the known interval is used to construct the grayscale prediction model; the historical voltage data obtained in the extrapolated interval is used for extrapolation prediction of the grayscale prediction model; Determining the historical voltage data according to the historical voltage data outputted by the proton exchange membrane fuel cell each time; The historical voltage data is denoised and outliers are removed to obtain the preprocessed data.
3. The method for predicting voltage decay of a proton exchange membrane fuel cell according to claim 1, characterized in that: The multiple intrinsic mode functions are subjected to frequency classification to obtain high-frequency sequences and low-frequency sequences, specifically including: According to the generation order of the plurality of intrinsic mode functions, taking the first generated intrinsic mode function as a starting point, i intrinsic mode functions are obtained; According to i intrinsic mode functions, i indicators are constructed; wherein the i-th indicator is the sum of i intrinsic mode functions; Calculate the mean of i indicators; The high-frequency sequence and the low-frequency sequence are determined based on the mean value and a first threshold.
4. The method for predicting voltage decay of a proton exchange membrane fuel cell according to claim 3, characterized in that: Based on the mean and the first threshold, determining the high-frequency sequence and the low-frequency sequence specifically includes: Determine whether the difference between the mean and 0 is greater than or equal to the first threshold, and obtain a first result; If the first result is yes, the first i-1 intrinsic mode functions are determined as high-frequency sequences, and the intrinsic mode functions after the i-1th intrinsic mode function are determined as low-frequency sequences; If the first result is no, return to the step of "obtaining i intrinsic mode functions in the order of generating the multiple intrinsic mode functions, taking the first generated intrinsic mode function as the starting point", and update i to i+1 until the difference between the mean and 0 is greater than or equal to the first threshold.
5. The method for predicting voltage decay of a proton exchange membrane fuel cell according to claim 1, characterized in that: The high frequency sequence and the low frequency sequence are reconstructed respectively to obtain a total high frequency mode and a total low frequency mode, specifically including: Adding each eigenmode function in the high frequency sequence to obtain the total high frequency mode; Each eigenmode function in the low-frequency sequence is added together to obtain the total low-frequency mode.
6. The method for predicting voltage decay of a proton exchange membrane fuel cell according to claim 2, characterized in that: Determining a fuel cell voltage decay prediction result according to the total high frequency mode, the total low frequency mode, the trend item and the grayscale prediction model specifically includes: Input the total high-frequency mode, the total low-frequency mode and the trend item as input data into the grayscale prediction model to obtain prediction data; the prediction data includes high-frequency prediction data, low-frequency prediction data and trend item prediction data; Calculating the difference between the predicted data and the input data to obtain a residual data sequence; Inputting the absolute value of each element in the residual data sequence into the grayscale prediction model to obtain a residual prediction value; Based on the known interval and the extrapolated interval, the residual prediction value is corrected to obtain a corrected residual prediction value; Adding the corrected residual prediction value to the prediction data to obtain final prediction data; The final prediction data is used as the fuel cell voltage decay prediction result.
7. The method for predicting voltage decay of a proton exchange membrane fuel cell according to claim 6, characterized in that: Based on the known interval and the extrapolated interval, the residual prediction value is corrected to obtain a corrected residual prediction value, specifically including: When the residual prediction value is within the known interval, the residual prediction value is corrected according to the positive or negative state of each element in the residual data sequence to determine the corrected residual prediction value; When the residual prediction value is located in the extrapolation interval, the residual prediction value is corrected by a Markov residual correction method to determine the corrected residual prediction value.
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 method for predicting voltage decay of 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 method for predicting voltage decay of a proton exchange membrane fuel cell according to any one of claims 1 to 7 is implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting voltage decay of a proton exchange membrane fuel cell according to any one of claims 1 to 7 is implemented.