Abstract: For the fault diagnosis of proton exchange membrane fuel cell（PEMFC）system,a fault diagnosis method based on P-L dual feature extraction was proposed. P-L dual feature extraction is used to extract features from the preprocessed sample data. Through redundant variable removal and secondary feature extraction,classification features are preserved to the maximum extent and the dimension of sample data is effectively reduced. Binary tree multi-class support vector machine and extreme learning machine are used to classify 2D fault feature vectors and realize fault diagnosis. Through the example verification,compared with the feature extraction effect of linear discriminant analysis,P-L dual feature extraction improves the diagnostic accuracy of the test set of the same classifier by21.19%,and the diagnostic accuracy reaches 99.27%,realizing the accurate and rapid diagnosis of membrane dry and hydrogen supply faults in PEMFC system.