Fault diagnosis method of proton exchange membrane fuel cell for tramcars

Abstract

The invention discloses a fault diagnosis method of a proton exchange membrane fuel cell for tramcars, comprising the following steps: collecting various types of data output during the actual operation of a fuel cell tramcar as original preparation data, and carrying out normalization and dimension reduction on the original preparation data to obtain original data; screening the original data based on fuzzy c-means clustering of a simulated annealing genetic algorithm to obtain sample data of fault diagnosis; and inputting the sample data to an established deep confidence fault diagnosis network based on an SMOTE algorithm, and outputting an optimal fault diagnosis accuracy and a classified fuel cell fault result. Accurate fault diagnosis data of a proton exchange membrane fuel cell for tramcars can be obtained. By processing the original data, a globally optimal solution can be found, more accurate sample data can be obtained, and the accuracy of model classification can be improved.By constructing the fault diagnosis network, unbalanced data can be processed, and the classification accuracy is greatly improved.

Description

technical field [0001] The invention belongs to the technical field of trams, in particular to a fault diagnosis method for proton exchange membrane fuel cells used in trams. Background technique [0002] Fuel cells are clean, efficient, and sustainable, and they are one of the most potential alternative energy sources in the 21st century. Among various fuel cells, proton exchange membrane fuel cells (PEMFC) also have high power density, fast response, With the advantages of low operating temperature and no electrolyte loss, it has broad application prospects in civil and military fields such as transportation, aerospace, and underwater vehicles. The fuel cell system for trams is divided into fuel cell modules, hydrogen supply system, air supply system, coolant circulation system, heat dissipation system and hydrogen circulation system, in which the performance of the fuel cell stack can usually reflect the status of the entire system. Since the fuel cell system is a very c...

AI Technical Summary

Problems solved by technology

However, the fault data of trams during operation is usually less than the normal data, and different fault data are usually unbalanced, especially the hydrogen leakage fault of the fuel cell stack of the tram, which is not only easy to interact with other faults Coupling occurs, and the fault data is relatively small. For such extremely unbalanced data, it is more difficult in the machine learning method, which will reduce the accuracy of the deep confidence network.

Images