A K-Means/Support Vector Machine Based SelfAdaptive Online Fault Diagnosis Method for Fuel Cell Systems

Abstract:

Based on k-means and support vector machine (SVM) algorithms, an online self-adaptive fault diagnosis method for automotive fuel cell system (FCS) is proposed. By continuously acquiring cell voltages and using k-means clustering to improve the original SVM classifier model, this method can achieve online self-adaption of the classifier. The experimental data from published papers were used to verify and analyze the results. The results show that the proposed method can effectively adjust the fault classifier online to detect the fault after changing the FCS system characteristics.