Sparse filtering based intelligent fault diagnosis using IPSO-SVM

Intelligent fault diagnosis has became a focus of fault diagnosis, which can quickly and efficiently process collected signals and obtain high accurate diagnosis results. Traditionally, intelligent fault diagnosis subjects to a lot of prior knowledge. In this work, a novel method based on the sparse features is proposed. The sparse features directly are extracted from raw mechanical vibration signals using unsupervised learning. Conventionally, Support Vector Machine optimized by Improved Particle Swarm (IPSO-SVM) is used to classify the health condition based on the sparse features for each sample. By introducing improved parameters into PSO algorithm, the global search ability of the PSO algorithm is enhanced. The proposed method is validated by a case of rolling element bearings datasets. In the experiments, we set up fifteen datasets to verify the proposed method. The results show that the proposed method can provide a simple and accurate fault diagnosis for bearing multi-faults.