Fault classification method based on random projections and SVM

SVM is an effective method for fault classification. However, it has computational problem while dealing with high dimensional datasets. To overcome this issue, we propose an efficient dimension reduction technique named random projection combined with the SVM for fault classification in this paper. Compared with well known dimension reduction method like PCA, it provides better tradeoff between pairwised distances reservation and computational time. The experiments on rolling bearing system demonstrate the feasibility and effectiveness of the proposed method.