CN116840722A - Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell - Google Patents

Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell Download PDF

Info

Publication number
CN116840722A
CN116840722A CN202310685096.9A CN202310685096A CN116840722A CN 116840722 A CN116840722 A CN 116840722A CN 202310685096 A CN202310685096 A CN 202310685096A CN 116840722 A CN116840722 A CN 116840722A
Authority
CN
China
Prior art keywords
fuel cell
exchange membrane
proton exchange
variable
membrane fuel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202310685096.9A
Other languages
Chinese (zh)
Other versions
CN116840722B (en
Inventor
王政
张楚
王业琴
王熠炜
陈佳雷
陈杰
彭甜
葛宜达
张学东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huaiyin Institute of Technology
Original Assignee
Huaiyin Institute of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huaiyin Institute of Technology filed Critical Huaiyin Institute of Technology
Priority to CN202310685096.9A priority Critical patent/CN116840722B/en
Publication of CN116840722A publication Critical patent/CN116840722A/en
Application granted granted Critical
Publication of CN116840722B publication Critical patent/CN116840722B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
    • G01R31/392—Determining battery ageing or deterioration, e.g. state of health
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02E60/30—Hydrogen technology
    • Y02E60/50—Fuel cells

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Human Resources & Organizations (AREA)
  • General Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Software Systems (AREA)
  • Computational Mathematics (AREA)
  • Strategic Management (AREA)
  • Pure & Applied Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Tourism & Hospitality (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • General Business, Economics & Management (AREA)
  • Databases & Information Systems (AREA)
  • Development Economics (AREA)
  • Operations Research (AREA)
  • Marketing (AREA)
  • Algebra (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Quality & Reliability (AREA)
  • Educational Administration (AREA)

Abstract

The invention discloses a performance degradation evaluation and life prediction method for a proton exchange membrane fuel cell, which comprises the steps of obtaining monitoring data of the proton exchange membrane fuel cell in advance, determining output variables of the monitoring data of the proton exchange membrane fuel cell by using a Pearson correlation analysis method, extracting and obtaining an input variable set by adopting MIC characteristics, and training by using a neural network model combined by an improved Ling-Ri search algorithm LGTSOA optimization time convolution network and a random vector network model RVFL to obtain a performance degradation evaluation prediction model LGTSOA-TCN-RVFL of the proton exchange membrane fuel cell, thereby realizing performance degradation evaluation and life prediction of the proton exchange membrane fuel cell. The invention provides more accurate prediction results for the performance degradation of the proton exchange membrane fuel cell and also provides more reliable basis for the life prediction of the proton exchange membrane fuel cell.

Description

Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell
Technical Field
The invention belongs to the technical field of fuel cells, and particularly relates to a performance degradation evaluation and life prediction method for a proton exchange membrane fuel cell.
Background
Currently, the global energy structure is mainly fossil energy, and other various forms of energy are auxiliary. However, the gradual depletion of fossil energy and the increasing environmental pollution problem, hydrogen energy is rising as a well-known clean low-carbon energy source in order to solve the energy shortage problem and to better save energy and reduce emission. Hydrogen is used as a clean energy carrier, has wide sources and wide application, can effectively reduce the specific gravity of fossil energy, improves the clean development level, and develops hydrogen energy as an important carrier for constructing a 'multi-energy complementary' energy supply system, thereby being an important power aid for realizing energy transformation and upgrading.
The hydrogen fuel cell (PEMFC) is used as an important tool for utilizing hydrogen energy, and is different from the traditional storage battery which provides electric energy in an energy storage mode, and the hydrogen fuel cell converts chemical energy into electric energy through electrochemical reaction between hydrogen and oxygen, so long as the hydrogen fuel cell has sufficient hydrogen source and air, the conversion process can be continued all the time, and zero pollution emission is achieved. Due to the characteristics of green environmental protection, low noise, high energy conversion efficiency and the like, the method has been widely applied to various fields. However, its durability significantly hinders the large-scale deployment and commercial development of hydrogen fuel cells.
The performance degradation estimation prediction of the hydrogen fuel cell is mainly to estimate the degradation state based on past operational summaries and current operation, and then to minimize maintenance costs and extend the remaining service life based on the state. Life prediction is a core technology in the process of fault prediction and health management, and by predicting the time at which a fault may occur, reliability is improved, which helps to monitor health and estimate the remaining life of a hydrogen fuel cell, providing a cost-effective strategy for scheduled maintenance. However, due to the nature of the hydrogen fuel cell itself, degradation is likely to occur during storage and operation, resulting in accelerated performance loss and shortened service life. The phenomenon of battery degradation can lead to the capacity reduction of the hydrogen fuel battery, so that the hydrogen fuel battery needs to be subjected to early warning evaluation, the time required by the hydrogen fuel battery to meet the performance standard of different loss degrees is required to be predicted, the performance degradation of the hydrogen fuel battery is evaluated, the life prediction value of the hydrogen fuel battery is obtained on the basis, and related devices are maintained, so that the safety accidents caused by the battery problem are avoided.
Disclosure of Invention
The invention aims to: aiming at the problems pointed out in the background art, the invention provides a performance degradation evaluation and life prediction method for a proton exchange membrane fuel cell, which realizes accurate and rapid prediction of the performance degradation evaluation and the residual life of the proton exchange membrane fuel cell.
The technical scheme is as follows: the invention provides a performance degradation evaluation and life prediction method of a proton exchange membrane fuel cell, which comprises the following steps:
(1) Monitoring data of a proton exchange membrane fuel cell are obtained in advance;
(2) Processing the data obtained in the step (1) by adopting a pearson correlation analysis method, and analyzing PEMFC health indexes;
(3) Calculating MIC values of variables and output variables in the monitoring data in the step (1) based on a maximum correlation coefficient method, wherein an input variable set is arranged in a descending order according to the MIC values, an optimal input variable set is selected, and a training set and a testing set are divided;
(4) Constructing a TCN-RVFL model, and optimizing the TCN-RVFL model by utilizing an improved Ling-Ri search algorithm LGTSOA;
(5) Inputting the training set obtained in the step (3) into the LGTSOA-TCN-RVFL for training, and predicting through the testing set to finally obtain the degradation trend of the health index and the service time of the fuel cell; obtaining a prediction result of fuel cell degradation evaluation;
(6) And determining an invalidation threshold according to health indexes representing the health condition of the PEMFC, and calculating the observed residual service life and the predicted residual service life under different invalidation thresholds.
Further, the monitoring data in step (1) includes aging time, cell and stack voltage, current density, hydrogen inlet and outlet temperatures, air inlet and outlet temperatures, cooling water inlet and outlet temperatures, hydrogen inlet and outlet pressures, air inlet and outlet pressures, hydrogen inlet and outlet flow rates, air inlet and outlet flow rates, cooling water flow rates, and cooling water flow rates.
Further, the step (2) is implemented by the following formula:
wherein R (X, Y) represents PCC between the variable X, i.e., the monitored data in the PEM fuel cell, and Y, i.e., the aging time of the PEM fuel cell, cov (X, Y) represents covariance between the variables X and Y,and->Represents the standard deviation of the variables X and Y, respectively, < >>And->Mean values of variables X and Y are represented respectively; let PCC between the first variable and the second variable be R 12 And so on, obtaining an m×n order PCC matrix R PCC The expression is as follows:
solving the correlation coefficient of the monitoring variable in the data set according to the formula (1) respectively, and outputting the correlation coefficient V of each fuel cell monitoring variable and the fuel cell aging time variable PCC The method comprises the steps of carrying out a first treatment on the surface of the And selecting the monitoring variable which is highly positively or negatively correlated with time according to the correlation coefficient result of each fuel cell monitoring variable and the fuel cell time variable, and determining the monitoring variable as a health index for representing the health condition of the PEMFC.
Further, the implementation process of the step (3) is as follows:
the correlation between the input variable and the output variable is measured based on the maximum information coefficient MIC, and a scattered point data set D= { (X) formed by X and Y is obtained i ,y i ) I=1, 2, …, n }, where x is all fuel cell monitoring data parameters except the set of output variables obtained above and y is the set of output variables obtained above; grid divided into a column b row, denoted g= (a, b), then MI value of D in G V mi (D| G ) Expressed as:
wherein P (x) and P (y) represent a fuel cell input variable set and a fuel cell inputThe edge probability density of the output variable set, P (x, y) represents the joint probability density of the fuel cell input variable set and the fuel cell output variable set; v (V) mi (D| G ) Selecting the maximum value, and normalizing to obtain MIC value V mic (D)：
Wherein f (n) =n 0.6 Representing an upper limit on the number of meshing grids, V mic (D) The larger the closer to 1, the stronger the correlation of the fuel cell input variable x and the output variable y; and selecting the first k input variables as the optimal input variable set according to the preset MIC value.
Further, the TCN-RVFL model construction process in the step (4) is as follows:
connecting data output by a space Dropout layer in a TCN model residual block with a hidden layer of an RVFL model as input of the RVFL model; in order to avoid the problem that the output of the TCN model is different from the input of the RVFL model hiding layer in dimension, a one-dimensional convolution kernel in the TCN model is utilized to change the size of an input channel so as to be added with the output, and the output of the TCN model is kept consistent with the input dimension of the RVFL model hiding layer.
Further, the improved Ling day search algorithm LGTSOA in the step (4) is a sine wave integrated with the Lev flight and gold, and the specific implementation process is as follows:
the LGTSOA implementation has five phases, namely a star system phase, a star phase, a sun-up phase, a planet phase and a development phase;
in the sun-up stage, the planet in the star system is fully optimized by utilizing Levy flight, the global searching capability of the star system is improved, and the star position formula is updated as follows:
D＝c 6 L S,i (14)
NS＝c 7 3 L S (15)
wherein L is S,new,i The updated star position is used for defining i star numbers according to the dimension of the input variable of the fuel cell, L S,i Is the position of the star when not updated; c 6 And c 7 Respectively a random number and a random vector, wherein alpha is a random step length;is dot multiplied; levy is a constraint under a random search path conforming to Levy distribution; mu and upsilon obey a normal distribution of standards; λ=1.5; />Searching for an amplification factor; Γ is a gamma function; calculating the updated brightness of the star and arranging the brightness in a descending order to divide the star grade;
in the development stage, the optimal planet position is determined by adding new knowledge and introducing a golden sine algorithm to traverse all values of a sine function, and meanwhile, the searching speed can be greatly improved by determining the position update of the planet so as to achieve good balance between searching and development, and the formula is as follows:
K＝c 11 z L c (20)
L best,Z ＝L new,Z |sin(R 1 )|+R 2 sin(R 1 )|x 1 P Z -x 2 L new,Z | (21)
wherein c k Representing different new knowledge conditions of addition, L best,Z Is the optimal planetary position; l (L) new,Z Obtaining planets for adding new knowledgeK is the expression mode of new knowledge, c 11 Is a random vector, L c =unirnd (l, h) is the upper and lower limits of the input variable set in proton exchange membrane fuel cell, R 1 And R is 2 Is a random number, R 1 Determining the distance of planet movement in the next iteration, R 1 ∈[0,2π]；R 2 Determining the position updating direction of the next iteration, R 2 ∈[0,π]；x 1 And x 2 The coefficient obtained by golden section reduces search space to lead the planet to approach the optimal value, namely the optimal planet position, as the optimal super parameter of the model TCN-RVFL, thereby reducing the error between the predicted value and the true value of the output variable of the proton exchange membrane fuel cell.
Further, the step (6) is implemented by the following sum formula:
health indexes for representing health conditions of the PEMFC, defining proper failure thresholds according to initial data of the health indexes, and calculating residual service life R of observation under different failure thresholds rul And predicted remaining useful life P rul ：
Wherein T is fred To predict the time of onset, T mFT For the time when the observed stack voltage first reaches the failure threshold, T fFT Is the time at which the predicted stack voltage first reaches the failure threshold.
The invention also provides an apparatus device comprising a memory and a processor, wherein:
a memory for storing a computer program capable of running on the processor;
and the processor is used for executing the steps of the proton exchange membrane fuel cell performance degradation evaluation and service life prediction method when the computer program is run.
The invention also provides a storage medium, wherein the storage medium is stored with a computer program, and the computer program realizes the steps of the proton exchange membrane fuel cell performance degradation evaluation and life prediction method when being executed by at least one processor.
The beneficial effects are that: compared with the prior art, the invention has the beneficial effects that: 1. the invention adopts the pearson correlation analysis method to preprocess the data, selects the health index which can represent the PEMFC, namely, the output variable set, and selects the parameter variable set with high correlation degree with the health index as the input variable set by the maximum information number MIC method on the basis; 2. based on the limitation of a time convolution network and a random vector network model, a data driving model of the time convolution network combined with a random vector network model (TCN-RVFL) is built, a Dropout layer regularized random vector network model of a residual error module in the time convolution network is utilized, namely, the output of the TCN is used as the input of the RVFL, a TCN-RVFL model is built, and the respective advantages of the two models are fully exerted; 3. the improved method is that a golden sine and Lev flight algorithm is introduced based on a Ling day search algorithm TSOA, a TCN-RVFL model is optimized by utilizing the improved Ling day search algorithm, an improved LGTSOA-TCN-RVFL model is established, degradation evaluation prediction of fuel cell health indexes is obtained, and the residual service lives of proton exchange membrane fuel cells are predicted by combining calculation of the residual service lives observed under different failure thresholds and the predicted residual service lives.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of data preprocessing;
FIG. 3 is a flow chart of a modified Ling Ri search algorithm;
figure 4 is a flow chart of TCN in combination with RVFL model.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings.
The invention provides a degradation evaluation and life prediction method of a proton exchange membrane fuel cell, which mainly comprises the steps of preprocessing data to select an input variable set and a health index output variable set, establishing a TCN-RVFL model, predicting degradation indexes of the fuel cell by combining an improved Ling day search algorithm, and realizing degradation evaluation and life prediction of the proton exchange membrane fuel cell by combining the observation of residual service life and the prediction of residual life; as shown in fig. 1, the method specifically comprises the following steps:
step 1: proton exchange membrane fuel cell monitoring data are collected, including aging time, cell and stack voltage, current density, hydrogen inlet and outlet temperatures, air inlet and outlet temperatures, cooling water inlet and outlet temperatures, hydrogen inlet and outlet pressures, air inlet and outlet pressures, hydrogen inlet and outlet flow rates, air inlet and outlet flow rates, cooling water flow rates, and cooling water flow rates.
Step 2: as shown in fig. 2, the data collected in the step 1 are processed by pearson correlation analysis, and selection of PEMFC health index (output variable) is analyzed.
The pearson correlation coefficient (Pearson Correlation Coefficient, PCC) describes the proportion of the total variance in the observed data that can be interpreted by a linear model of the study variable. The larger the absolute value, the better the correlation between the variables. The calculation formula of PCC is as follows:
wherein R (X, Y) represents PCC between the variable X (i.e., the parameter of the monitored data in the PEM fuel cell) and Y (i.e., the aging time of the PEM fuel cell), cov (X, Y) represents covariance between the variables X and Y,andrepresents the standard deviation of the variables X and Y, respectively, < >>And->Representing the mean of variables X and Y, respectively. Let PCC between the first variable and the second variable be R 12 And so on, finally obtainmXn order PCC matrix R PCC The expression is as follows:
solving correlation coefficients of the monitoring variables in the data set according to a PCC (policy and charging control) calculation formula (1) respectively, and outputting the monitoring variables of each fuel cell and the aging V of the fuel cell PCC . And selecting the monitoring variables which are highly positive (negative) correlated with time according to the correlation coefficient results of the monitoring variables and the time variables, and determining the monitoring variables as health indexes (output variables) for representing the health condition of the PEMFC.
Step 3: as shown in fig. 2, after determining the health index (output variable) according to pearson correlation analysis, the influence of other physical monitoring variables on the output variable of the PEMFC needs to be further considered, in order to obtain the key influencing variable containing the degradation information of the PEMFC, the modeling difficulty of the subsequent life prediction work is reduced, and the modeling based on the maximum information coefficient (Maximum information coefficient, MIC) is adopted. The MIC is used for measuring the correlation between the input variable and the output variable, and has the advantages of low computational complexity, high robustness and the like.
MIC is an efficient tool to calculate the linear (or nonlinear) intensity between two variables, the main idea of which is: if there is a relationship between the two variables, meshing in the scatter plot of the two variables, calculating mutual information (Mutual Information, MI) under different partitioning schemes, and normalizing the maximum MI value in all the partitioning schemes to obtain the MIC. Scattered point data set d= { (X) composed of variables X and Y i ,y i ) I=1, 2, …, n }, where x is all fuel cell monitoring data parameters except the set of output variables obtained above and y is the set of output variables obtained above; grid divided into a columns and b rows, wherein one division scheme is denoted as g= (a, b), then the MI value V of D in G mi (D| G ) Can be expressed as:
wherein P (x) and P (y) represent fuelThe edge probability densities of the input and output variable sets of the cell, P (x, y) represent the joint probability densities of the input and output variable sets of the fuel cell. V in partitioning scheme mi (D| G ) Selecting the maximum value, and normalizing to obtain MIC value V mic (D) The calculation formula is as follows:
wherein f (n) =n 0.6 Representing an upper limit on the number of meshing grids, V mic (D) The larger the closer to 1, the stronger the correlation of the fuel cell input variable X and the output variable Y. And selecting the first k input variables as the optimal input variable set according to the preset MIC value.
Step 4: and constructing a TCN-RVFL model, and optimizing the TCN-RVFL model by utilizing an improved Ling-Ri search algorithm LGTSOA. As shown in fig. 3, the improved linger search algorithm LGTSOA is a combination of the lewy flight and the golden sine, and the flow is as follows:
LGTSOA implements a total of five phases, namely a star phase, a sun phase, a planet phase, and a development phase.
Stage of the star: initializing the center L of the galaxy c Then determining the habitable area L in the star system R,n The case of finding the best sidereal system in habitable areas. And finally selecting the star with the best suitability.
L R,n ＝L c +D-NS (5)
Wherein L is c =unified (l, h) is the upper and lower limits of the input variable set in the fuel cell, and n is the number of star systems, i.e. the dimension of the input variable set of the proton exchange membrane fuel cell.
Wherein c 1 A random number of 0 to 1. D is the range of positions where the star system appears in both active and inactive situations, i.e. the range defining the position of the set of search input variablesAn area.
NS＝(c 2 ) 3 unifrnd(l,h) (7)
Wherein NS is signal noise generated during observation, C 2 Is a random vector.
Star stage: in the above-mentioned range of finding star systems, a star corresponding to the star system is selected, and the formula is as follows:
L S,i ＝L R,i +D-NS (8)
NS＝(c 5 ) 3 unifrnd(l,h) (10)
wherein L is S,i Indicating the position of the star L R,i Is the ith star system in the star system stage, c 3 And c 4 A random number of 0 to 1, c 5 Is a random vector number.
Setting an optimized fitness function:wherein->The jth predicted value, y, in the input variable for the nth proton exchange membrane fuel cell n,j And calculating the fitness value of each star and planet for the j-th true value in the n-th proton exchange membrane fuel cell input variable, namely the deviation between the predicted value and the actual value of the fuel cell input variable set, so as to obtain the optimal star and planet position, wherein the star and the planet with lower fitness values represent smaller deviation between the predicted value and the actual value of the fuel cell.
The sun-up stage: in order to determine whether there are any planets in the star, the amount of light received from the star is measured by detecting the sun-up phenomenon, and whether there are any planets passing. The luminance of a star is its intrinsic luminance, which is related to the energy radiated by the star per second, so the luminance received from the star is determined and updated as follows:
wherein L is i And R is i The brightness and brightness level of the ith fixed star, d i Indicating the distance between telescope and ith fixed star, telescope L T Is random and does not change during the optimization process. The range of exploring the star system is enlarged by utilizing the mutation characteristic of Levy flight, the planets in the star system are fully optimized, the global searching capability of the star system is greatly improved, and the position formula of the updated star is as follows:
D＝c 6 L S,i (14)
NS＝c 7 3 L S (15)
wherein L is S,new,i Is the updated star position, i stars are defined according to the dimension of the fuel cell input variable, L S,i Is the position of the star when not updated; c 6 And c 7 Respectively a random number and a random variable, wherein alpha is a random step length;is dot multiplied; levy is a constraint under a random search path conforming to Levy distribution; mu and upsilon obey a normal distribution of standards; λ=1.5; />Searching for an amplification factor; Γ is a gamma function; and calculating the updated brightness of the star and arranging the star grades in a descending order, wherein the calculation formula is the formula (11).
And (3) planetary stage: the initial position L of the detected planet is determined by first determining the initial position of the detected planet in view of the fact that the light received by the telescope comes from the star, so that when the planet passes between the star and the telescope, the brightness of the light is reduced, i.e. the sun is in the sun Z The planetary condition of the current position between the star and the telescope can be determined by using the average value of the two relative positions of the star and the telescope, and the formula is as follows:
wherein c 8 Is a random number.
Development stage: the optimal planet is determined for each star, the characteristics of the planet and the conditions of inoculation life are found at the stage, all values of a sine function can be traversed by adding new knowledge and introducing a golden sine algorithm, meanwhile, the searching speed can be greatly improved by determining the position update of the planet, so that searching and developing are well balanced, the optimal planet position is determined next time, and the formula is as follows:
K＝c 11 z L c (20)
L best,Z ＝L new,Z |sin(R 1 )|+R 2 sin(R 1 )|x 1 P Z -x 2 L new,Z | (21)
wherein c k Representing different new knowledge situations added; l (L) best,Z Is the optimal planetary position; l (L) new,Z To add new knowledge to get the position update of the planet, K is the expression mode of the new knowledge, c 11 Is a random vector, L c =unirnd (l, h) is proton exchange membraneUpper and lower limits of input variable set in fuel cell, R 1 And R is 2 Is a random number, R 1 Determining the distance of planet movement in the next iteration, R 1 ∈[0,2π]；R 2 Determining the position updating direction of the next iteration, R 2 ∈[0,π]；x 1 And x 2 The coefficient obtained by golden section reduces search space to lead the planet to approach the optimal value, namely the optimal planet position, as the optimal super parameter of the model TCN-RVFL, thereby reducing the error between the predicted value and the true value of the output variable of the proton exchange membrane fuel cell.
As shown in fig. 4, a TCN-RVFL model is established, as follows:
the TCN network is stacked by several residual blocks. The internal construction of the residual block includes an extended causal convolutional layer, a weight normalization operation, an activation function ReLU, a space Dropout, residual chaining, and an optional one-dimensional convolutional layer. Thus, there is excellent performance in processing the fuel cell multi-dimensional input variable set, but the multi-layered residual block requires a lot of time to train the fuel cell input variable set.
The RVFL model belongs to a typical single hidden layer feedforward neural network, and has the characteristics of small required training sample, high training speed, higher analysis precision and the like due to the direct connection design of the RVFL model, besides an input layer, an hidden layer and an output layer, a linear link is also arranged to directly connect the input layer and the output layer. However, due to random generation of network part parameters and structural risk of the network model, when the RVFL model faces to the condition that the working condition of the fuel cell is complex and changeable, hidden layer nodes of the simulated sample points are greatly increased, a good training effect can be achieved, and an overfitting condition such as poor prediction result of performance degradation indexes of the fuel cell can be achieved.
It is therefore proposed herein that the TCN model in combination with the RVFL model has the problems associated with the multi-dimensional and data-intensive nature of the fuel cell input variable set. The spatial Dropout in the TCN is a regularization technology, which can reduce the risk of over fitting of the neural network, so that the data output by the spatial Dropout layer in the residual block of the TCN model is used as the input of the RVFL model to be connected with the hidden layer of the RVFL model, namely, the RVFL model is used for replacing the output layer of the TCN. In order to avoid the problem that the output of the TCN model is different from the input of the RVFL model hiding layer in dimension, a one-dimensional convolution kernel in the TCN model is utilized to change the size of an input channel so as to be added with the output, and the output of the TCN model is kept consistent with the input dimension of the RVFL model hiding layer.
The TCN and RVFL neural network are combined, the time sequence problem is solved more effectively through the TCN model, the information processing capability of the memory unit is improved greatly when the TCN model is entered into the RVFL model, and therefore the prediction model can master the complexity and the correlation of the time sequence more efficiently.
Step 5: inputting the training set obtained in the step 3 into the LGTSOA-TCN-RVFL for training, and predicting through the test set to finally obtain the degradation trend of the health index and the service time of the fuel cell; and obtaining a prediction result of fuel cell degradation evaluation:
the improved Lingri search algorithm optimizing time convolution network combines the random vector model LGTSOA-TCN-RVFL to have better robustness on the health index prediction of the fuel cell, can avoid the over-fitting phenomenon, increases the accuracy of the prediction, and provides a favorable foundation for the life prediction of the fuel cell.
And 6, determining an invalidation threshold according to health indexes representing the health condition of the PEMFC, and calculating the observed residual service life and the predicted residual service life under different invalidation thresholds. The implementation process is as follows:
health indicators (output variable sets) characterizing the health of the PEMFC, defining suitable failure thresholds from initial data of the health indicators, and calculating observed remaining useful life (R rul ) And predicted remaining useful life (P rul ) The calculation formula is as follows:
wherein T is fred To predict the time of onset, T mFT For the time when the observed stack voltage first reaches the failure threshold, T fFT Is the time at which the predicted stack voltage first reaches the failure threshold.
Predictive evaluation index: to evaluate the model's effect on the PEMFC stack output voltage degradation trend prediction, a Root Mean Square Error (RMSE), a Mean Absolute Percentage Error (MAPE) and a R square (R 2 ) And (5) an index. Wherein smaller RMSE and MAPE indicate that the predicted value is closer to the observed value; r is R 2 The closer to 1, the better the fitting effect of the model. The calculation formulas of the three indexes are as follows:
where M is the predicted value of the fuel cell output variable set, i is the ith sample, F is the actual value of the fuel cell output variable set, and N is the total number of samples.
The invention also provides an apparatus device comprising a memory and a processor, wherein: a memory for storing a computer program capable of running on the processor; and the processor is used for executing the steps of the proton exchange membrane fuel cell performance degradation evaluation and service life prediction method when the computer program is run.
The invention also provides a storage medium, and a computer program is stored on the storage medium, and the computer program realizes the steps of the proton exchange membrane fuel cell performance degradation evaluation and life prediction method when being executed by at least one processor.
The foregoing embodiments are merely illustrative of the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the present invention and to implement the same, not to limit the scope of the present invention. All equivalent changes or modifications made according to the spirit of the present invention should be included in the scope of the present invention.

Claims (9)

1. The performance degradation evaluation and life prediction method for the proton exchange membrane fuel cell is characterized by comprising the following steps of:
(1) Monitoring data of a proton exchange membrane fuel cell are obtained in advance;
(2) Processing the data obtained in the step (1) by adopting a pearson correlation analysis method, and analyzing PEMFC health indexes;
(3) Calculating MIC values of variables and output variables in the monitoring data in the step (1) based on a maximum correlation coefficient method, wherein an input variable set is arranged in a descending order according to the MIC values, an optimal input variable set is selected, and a training set and a testing set are divided;
(4) Constructing a TCN-RVFL model, and optimizing the TCN-RVFL model by utilizing an improved Ling-Ri search algorithm LGTSOA;
(5) Inputting the training set obtained in the step (3) into the LGTSOA-TCN-RVFL for training, and predicting through the testing set to finally obtain the degradation trend of the health index and the service time of the fuel cell; obtaining a prediction result of fuel cell degradation evaluation;
(6) And determining an invalidation threshold according to health indexes representing the health condition of the PEMFC, and calculating the observed residual service life and the predicted residual service life under different invalidation thresholds.
2. The method of claim 1, wherein the monitoring data in step (1) includes aging time, cell and stack voltage, current density, hydrogen inlet and outlet temperatures, air inlet and outlet temperatures, cooling water inlet and outlet temperatures, hydrogen inlet and outlet pressures, air inlet and outlet pressures, hydrogen inlet and outlet flow rates, air inlet and outlet flow rates, cooling water flow rates.
3. The method for evaluating the performance degradation and predicting the life of a proton exchange membrane fuel cell according to claim 1, wherein the step (2) is realized by the following formula:
wherein R (X, Y) represents PCC between the variable X, i.e., the monitored data in the PEM fuel cell, and Y, i.e., the aging time of the PEM fuel cell, cov (X, Y) represents covariance between the variables X and Y,and->Represents the standard deviation of the variables X and Y, respectively, < >>And->Mean values of variables X and Y are represented respectively; let PCC between the first variable and the second variable be R 12 And so on, obtaining an m×n order PCC matrix R PCC The expression is as follows:
solving the correlation coefficient of the monitoring variable in the data set according to the formula (1) respectively, and outputting the correlation coefficient V of each fuel cell monitoring variable and the fuel cell aging time variable PCC The method comprises the steps of carrying out a first treatment on the surface of the And selecting the monitoring variable which is highly positively or negatively correlated with time according to the correlation coefficient result of each fuel cell monitoring variable and the fuel cell time variable, and determining the monitoring variable as a health index for representing the health condition of the PEMFC.
4. The method for evaluating the performance degradation and predicting the service life of a proton exchange membrane fuel cell according to claim 1, wherein the implementation process of the step (3) is as follows:
the correlation between the input variable and the output variable is measured based on the maximum information coefficient MIC, and a scattered point data set D= { (X) formed by X and Y is obtained i ,y i ) I=1, 2, …, n }, where x is all fuel cell monitoring data parameters except the set of output variables obtained above and y is the set of output variables obtained above; grid divided into a column b row, denoted g= (a, b), then MI value of D in G V mi (D| G ) Expressed as:
wherein P (x) and P (y) represent the edge probability densities of the fuel cell input variable set and the fuel cell output variable set, and P (x, y) represents the joint probability densities of the fuel cell input variable set and the fuel cell output variable set; v (V) mi (D| G ) Selecting the maximum value, and normalizing to obtain MIC value V mic (D)：
Wherein f (n) =n 0.6 Representing an upper limit on the number of meshing grids, V mic (D) The larger the closer to 1, the stronger the correlation of the fuel cell input variable x and the output variable y; and selecting the first k input variables as the optimal input variable set according to the preset MIC value.
5. The method for evaluating performance degradation and predicting service life of a proton exchange membrane fuel cell according to claim 1, wherein the TCN-RVFL model construction process in step (4) is as follows:
connecting data output by a space Dropout layer in a TCN model residual block with a hidden layer of an RVFL model as input of the RVFL model; in order to avoid the problem that the output of the TCN model is different from the input of the RVFL model hiding layer in dimension, a one-dimensional convolution kernel in the TCN model is utilized to change the size of an input channel so as to be added with the output, and the output of the TCN model is kept consistent with the input dimension of the RVFL model hiding layer.
6. The method for evaluating performance degradation and predicting service life of a proton exchange membrane fuel cell according to claim 1, wherein the improved ling day search algorithm LGTSOA in the step (4) is a combination of a lewy flight and a golden sine, and comprises the following specific implementation processes:
the LGTSOA implementation has five phases, namely a star system phase, a star phase, a sun-up phase, a planet phase and a development phase;
in the sun-up stage, the planet in the star system is fully optimized by utilizing Levy flight, the global searching capability of the star system is improved, and the star position formula is updated as follows:
D＝c 6 L S,i (14)
NS＝c 7 3 L S (15)
wherein L is S,new,i The updated star position is used for defining i star numbers according to the dimension of the input variable of the fuel cell, L S,i Is the position of the star when not updated; c 6 And c 7 Respectively a random number and a random vector, wherein alpha is a random step length;is dot multiplied; levy is a constraint under a random search path conforming to Levy distribution; mu and upsilon obey a normal distribution of standards; λ=1.5; />Searching for an amplification factor; Γ is a gamma function; calculating the updated brightness of the star and arranging the brightness in a descending order to divide the star grade;
in the development stage, the optimal planet position is determined by adding new knowledge and introducing a golden sine algorithm to traverse all values of a sine function, and meanwhile, the searching speed can be greatly improved by determining the position update of the planet so as to achieve good balance between searching and development, and the formula is as follows:
K＝c 11 z L c (20)
L best,Z ＝L new,Z |sin(R 1 )|+R 2 sin(R 1 )|x 1 P Z -x 2 L new,Z | (21)
wherein c k Representing different new knowledge conditions of addition, L best,Z Is the optimal planetary position; l (L) new,Z To add new knowledge to get the position update of the planet, K is the expression mode of the new knowledge, c 11 Is a random vector, L c =unirnd (l, h) is the upper and lower limits of the input variable set in proton exchange membrane fuel cell, R 1 And R is 2 Is a random number, R 1 Determining the distance of planet movement in the next iteration, R 1 ∈[0,2π]；R 2 Determining the position updating direction of the next iteration, R 2 ∈[0,π]；x 1 And x 2 The coefficient obtained by golden section reduces search space to lead the planet to approach the optimal value, namely the optimal planet position, as the optimal super parameter of the model TCN-RVFL, thereby reducing the error between the predicted value and the true value of the output variable of the proton exchange membrane fuel cell.
7. The method for evaluating the performance degradation and predicting the life of a proton exchange membrane fuel cell according to claim 1, wherein the step (6) is realized by the following sum formula:
health indexes for representing health conditions of the PEMFC, defining proper failure thresholds according to initial data of the health indexes, and calculating residual service life R of observation under different failure thresholds rul And predicted remaining useful life P rul ：
Wherein T is fred To predict the time of onset, T mFT For the time when the observed stack voltage first reaches the failure threshold, T fFT Is the time at which the predicted stack voltage first reaches the failure threshold.
8. An apparatus device comprising a memory and a processor, wherein:
a memory for storing a computer program capable of running on the processor;
a processor for performing the steps of the proton exchange membrane fuel cell performance degradation assessment and lifetime prediction method as claimed in any one of claims 1-7 when running said computer program.
9. A storage medium having stored thereon a computer program which, when executed by at least one processor, implements the steps of the proton exchange membrane fuel cell performance degradation assessment and lifetime prediction method as claimed in any one of claims 1-7.
CN202310685096.9A 2023-06-09 2023-06-09 Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell Active CN116840722B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310685096.9A CN116840722B (en) 2023-06-09 2023-06-09 Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310685096.9A CN116840722B (en) 2023-06-09 2023-06-09 Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell

Publications (2)

Publication Number Publication Date
CN116840722A true CN116840722A (en) 2023-10-03
CN116840722B CN116840722B (en) 2024-02-23

Family

ID=88169721

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310685096.9A Active CN116840722B (en) 2023-06-09 2023-06-09 Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell

Country Status (1)

Country Link
CN (1) CN116840722B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118312737A (en) * 2024-04-08 2024-07-09 淮阴工学院 Proton exchange membrane fuel cell performance degradation interval prediction method

Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2947957A1 (en) * 2009-07-09 2011-01-14 Commissariat Energie Atomique METHOD AND DEVICE FOR INCREASING THE LIFETIME OF A PROTON EXCHANGE MEMBRANE FUEL CELL
US20130252116A1 (en) * 2012-03-21 2013-09-26 GM Global Technology Operations LLC Model Based Approach For In-Situ WVTD Degradation Detection In Fuel Cell Vehicles
CN108226809A (en) * 2018-04-13 2018-06-29 淮阴工学院 A kind of multi-model and battery SOC evaluation method
CN109324291A (en) * 2018-08-21 2019-02-12 西南交通大学 A kind of prediction technique for Proton Exchange Membrane Fuel Cells life prediction
EP3654047A1 (en) * 2018-11-14 2020-05-20 Commissariat à l'Énergie Atomique et aux Énergies Alternatives Diagnostic method for a fuel cell
CN111310387A (en) * 2020-02-19 2020-06-19 电子科技大学 Fuel cell life prediction method
CN111339712A (en) * 2020-02-24 2020-06-26 电子科技大学 Method for predicting residual life of proton exchange membrane fuel cell
CN112986827A (en) * 2021-04-12 2021-06-18 山东凯格瑞森能源科技有限公司 Fuel cell residual life prediction method based on deep learning
CN113657024A (en) * 2021-07-22 2021-11-16 武汉理工大学 Fuel cell life prediction method
CN114814589A (en) * 2022-01-20 2022-07-29 北京科技大学 Method and device for predicting remaining service life of PEMFC
CN114899457A (en) * 2022-05-23 2022-08-12 淮阴工学院 Fault detection method for proton exchange membrane fuel cell system
CN114943372A (en) * 2022-05-09 2022-08-26 国网浙江省电力有限公司宁波供电公司 Method and device for predicting life of proton exchange membrane based on Bayesian recurrent neural network
KR20230036776A (en) * 2021-09-08 2023-03-15 울산대학교 산학협력단 System and method for fault diagnosis of fuel cell energy management system based on digital twin
CN116047303A (en) * 2022-11-24 2023-05-02 海卓动力(青岛)能源科技有限公司 Method for comprehensively analyzing uniformity of commercial-size proton exchange membrane fuel cell

Patent Citations (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2947957A1 (en) * 2009-07-09 2011-01-14 Commissariat Energie Atomique METHOD AND DEVICE FOR INCREASING THE LIFETIME OF A PROTON EXCHANGE MEMBRANE FUEL CELL
US20130252116A1 (en) * 2012-03-21 2013-09-26 GM Global Technology Operations LLC Model Based Approach For In-Situ WVTD Degradation Detection In Fuel Cell Vehicles
CN108226809A (en) * 2018-04-13 2018-06-29 淮阴工学院 A kind of multi-model and battery SOC evaluation method
CN109324291A (en) * 2018-08-21 2019-02-12 西南交通大学 A kind of prediction technique for Proton Exchange Membrane Fuel Cells life prediction
EP3654047A1 (en) * 2018-11-14 2020-05-20 Commissariat à l'Énergie Atomique et aux Énergies Alternatives Diagnostic method for a fuel cell
CN111310387A (en) * 2020-02-19 2020-06-19 电子科技大学 Fuel cell life prediction method
CN111339712A (en) * 2020-02-24 2020-06-26 电子科技大学 Method for predicting residual life of proton exchange membrane fuel cell
CN112986827A (en) * 2021-04-12 2021-06-18 山东凯格瑞森能源科技有限公司 Fuel cell residual life prediction method based on deep learning
CN113657024A (en) * 2021-07-22 2021-11-16 武汉理工大学 Fuel cell life prediction method
KR20230036776A (en) * 2021-09-08 2023-03-15 울산대학교 산학협력단 System and method for fault diagnosis of fuel cell energy management system based on digital twin
CN114814589A (en) * 2022-01-20 2022-07-29 北京科技大学 Method and device for predicting remaining service life of PEMFC
CN114943372A (en) * 2022-05-09 2022-08-26 国网浙江省电力有限公司宁波供电公司 Method and device for predicting life of proton exchange membrane based on Bayesian recurrent neural network
CN114899457A (en) * 2022-05-23 2022-08-12 淮阴工学院 Fault detection method for proton exchange membrane fuel cell system
CN116047303A (en) * 2022-11-24 2023-05-02 海卓动力(青岛)能源科技有限公司 Method for comprehensively analyzing uniformity of commercial-size proton exchange membrane fuel cell

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
SAAD S. KHAN 等: "Novel dynamic semiempirical proton exchange membrane fuel cell model incoporating component voltages", 《INTERNATIONAL JOURNAL OF ENERGY RESEARCH》, vol. 42, no. 8, pages 2615 - 2630, XP071644895, DOI: 10.1002/er.4038 *
胡鹏飞: "质子交换膜燃料电池的健康度和剩余使用寿命的预测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》, no. 02, pages 015 - 1282 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118312737A (en) * 2024-04-08 2024-07-09 淮阴工学院 Proton exchange membrane fuel cell performance degradation interval prediction method
CN118312737B (en) * 2024-04-08 2024-09-20 淮阴工学院 Proton exchange membrane fuel cell performance degradation interval prediction method

Also Published As

Publication number Publication date
CN116840722B (en) 2024-02-23

Similar Documents

Publication Publication Date Title
Pan et al. Performance degradation prediction of proton exchange membrane fuel cell using a hybrid prognostic approach
CN110579709B (en) Fault diagnosis method for proton exchange membrane fuel cell for tramcar
Chen et al. Aging prognosis model of proton exchange membrane fuel cell in different operating conditions
CN117713221B (en) Micro-inversion photovoltaic grid-connected optimization system
CN116840722B (en) Performance degradation evaluation and life prediction method for proton exchange membrane fuel cell
Pang et al. Intelligent health states recognition of fuel cell by cell voltage consistency under typical operating parameters
CN116306226B (en) Fuel cell performance degradation prediction method
Wang et al. Parameters exploration of SOFC for dynamic simulation using adaptive chaotic grey wolf optimization algorithm
Huang et al. Correlation analysis and prediction of PEM fuel cell voltage during start-stop operation based on real-world driving data
CN115618732A (en) Nuclear reactor digital twin key parameter autonomous optimization data inversion method
Hongwei et al. An interpretable data-driven method for degradation prediction of proton exchange membrane fuel cells based on temporal fusion transformer and covariates
Li et al. Degradation prediction of proton exchange membrane fuel cell based on the multi-inputs Bi-directional long short-term memory
CN117829822B (en) Power transformer fault early warning method and system
Zhang et al. Intelligent diagnosis of proton exchange membrane fuel cell water states based on flooding-specificity experiment and deep learning method
Wang et al. Tracking and predicting technological knowledge interactions between artificial intelligence and wind power: Multimethod patent analysis
CN109978252A (en) A kind of high-power integrated fuel cell system running state prediction and evaluation method
CN114742363A (en) Energy efficiency state evaluation method, system and medium for wind turbine generator
Bai et al. Towards trustworthy remaining useful life prediction through multi-source information fusion and a novel LSTM-DAU model
CN113469420A (en) Electric power energy structure evaluation optimization method for multi-element power supply system
Xu et al. Research on Load Forecasting Based on CNN-LSTM Hybrid Deep Learning Model
CN116500480A (en) Intelligent battery health monitoring method based on feature transfer learning hybrid model
Liu et al. FSR‐SSL: A fault sample rebalancing framework based on semi‐supervised learning for PV fault diagnosis
Huang et al. Long-Term Health State Estimation of Energy Storage Lithium-Ion Battery Packs
Wang Remaining Useful Life Prediction of Proton Exchange Membrane Fuel Cell Based on Deep Learning
Jiurong et al. A data-driven fault diagnosis method for photovoltaic modules

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant