An On-line Fault Diagnosis Method for Gas Engine Using AP Clustering

摘要

The purpose of the paper is to implement an online fault diagnosis system for gas engine based on its data stream. The method proposed consists of three steps. First, we extract the features in a sliding window and reduce the dimension by PCA (Principal Components Analysis); second, the representative features are obtained by an AP (Affinity Propagation) clustering algorithm; finally, the fault pattern recognition is accomplished according to the distance between feature samples. The diagnosis results for three common faults show that this method achieves better performance than existing methods based on different clustering algorithms.