Motor Bearing Fault Diagnosis Based on Wavelet Packet Analysis and Sparse Filtering

With the development of artificial intelligence industry, it is more and more popular to combine deep learning technology with machine fault diagnosis recent years. The process of fault diagnosis includes two main steps: extracting fault features from signals and classify the health conditions based on the features. In traditional diagnosis methods, features extraction depends on human labor and prior knowledge so much, which makes the processes of fault diagnosis not only time-consuming and laborious, but also subjective and limited. According to the theory of unsupervised learning, a method for intelligent diagnosis of motor bearing is proposed in this paper. Firstly, wavelet packet analysis(WPA) is used to reduce noise by decomposing and reconstructing the original signal. Then, sparse filtering network is used to extract features from decomposed signals. Finally, according to the extracted features, support vector machine is used to classify the health conditions. The results show that the proposed method is more accurate than other related methods. Because of unsupervised learning, the proposed method depends on human labor and prior knowledge rarely.