Multi-source feature extraction of rolling bearing compression measurement signal based on independent component analysis

Abstract With the update of the sampling rate, automation and computation, data volume is increasing, it's critical to reduce the burden on the real-time data processing and remote diagnostics. In this paper, a composite fault diagnosis method of rolling bearing based on compressed sensing (CS) framework is proposed. Firstly, the influence of measurement matrix on the gaussianity of signal was analyzed, and Hadamard measurement matrix was used to compress and collect data. Then independent component analysis (ICA) was used to process the data collected by compression, and the data was separated and transformed based on the statistical independence. Finally, the reconstructed signal was analyzed by envelope spectrum, and the characteristic frequency of compound faults signal was extracted for fault diagnosis. The experiment result shows that the method can improve the reconstruction precision and the separation stability of fault signal and can effectively extract fault characteristics and realize fault diagnosis.