Study On Life Prediction Method Of Proton Exchange Membrane Fuel Cell Under Multiple Working Conditions

Life prediction is an important tool in the life prediction and health management technology of proton exchange membrane fuel cell(PEMFC).In this paper,the prediction accuracy,robustness and universality of the PEMFC life prediction algorithm under multiple working conditions are studied.Considering the influence of reversible recession on the life prediction algorithm,a hybrid prediction method is proposed to accurately predict the end point of PEMFC by monitoring the performance degradation process during operation.It has practical significance and application prospect.Firstly,this paper reviews the research status of PEMFC life prediction,introduces the working principle of PEMFC,analyzes the factors that influence the performance degradation.Based on the test platform of water-cooled hydrogen-air fuel cell built in the laboratory,the whole working condition PEMFC durability experiment is designed,providing theoretical basis and data support for the subsequent life prediction and evaluation.Secondly,by analyzing the characteristics of PEMFC degenerate data and the advantages and disadvantages of the traditional prediction algorithm,discrete wavelet transform and selfadaptive differential evolution algorithm are introduced to optimize the extreme learning machine,and the accuracy of the algorithm is improved.In view of three types of different working conditions,the simulation and comparative analysis are carried out based on the MATLAB simulation platform.The proposed algorithm has small time error,high prediction accuracy and strong robustness.Thirdly,in view of the voltage reversible recession caused by periodic performance testing problems,this paper sets up an equivalent circuit model of PEMFC.ZView software is used to predict the changes of internal parameters of the model,and the measured electrochemical impedance spectra at different times are fitted.The degradation of the PEMFC stack and the possible aging reasons are analyzed for three working conditions.The trend of voltage change with respect to time can be obtained by quadratic function fitting,and the voltage drop rate,namely the derivative of the quadratic fitting function,is taken as the slope of each reversible degradation.The reversible voltage obtained is a piecewise linear functionFinally,this paper proposes a hybrid prediction method based on model driven and data driven algorithm.The measured operating data and the updated estimated voltage are introduced as the reference values for the life evaluation of PEMFC in the next cycle.By modifying the voltage degradation rate,the data-driven prediction algorithm is adjusted and corrected,which can avoid the inaccurate prediction of the end of life due to the reversible voltage degradation.The proposed hybrid prediction algorithm retains the advantages of datadriven algorithm,and can track the trend of data more accurately.In dealing with the problem of reversible voltage degradation,the prediction accuracy of hybrid algorithm is improved,and the universality and practicability of the algorithm are verified.