A FAULT DIAGNOSIS METHOD OF ENVIRONMENT-FRIENDLY PROTON EXCHANGE MEMBRANE FUEL CELL FOR VEHICLES USING DEEP LEARNING

Proton Exchange Membrane Fuel Cell (PEMFC) has the advantages of high-power density, low operating temperature, high energy conversion rate and environmental friendliness. However, problems such as high maintenance cost, short service life and fast performance degradation restrict its commercial development. Aiming at the problem of fault classification and diagnosis of PEMFC system, a PEMFC fault diagnosis method based on Empirical Mode Decomposition (EMD) and improved Long Short-Term Memory (LSTM) is proposed. This paper uses EMD method to process the collected data, improve the effectiveness of fault information, and provide complete and effective data support for subsequent model learning and training; Based on the respective characteristics of LSTM network model and Recurrent Neural Network (RNN) model, a PEMFC fault determination method based on deep learning is proposed. In order to obtain a classifier with higher accuracy, improve the fault diagnosis accuracy of the PEMFC system, and ensure the normal and stable operation of the high-power PEMFC system for vehicles. The simulation experiment results show that the proposed method can effectively determine PEMFC faults, and the accuracy of system fault classification judgment reaches 98.23%, showing good feasibility and practicability, which will be very good for protecting the natural environment.