A rolling bearing fault diagnosis method based on Compressive sensing and Local characteristic-scale decomposition

Rolling bearings are an important part of the system with rotating parts. In the past, the rolling bearing fault diagnosis was based on the envelope of the bearing vibration waveform and FFT analysis to identify the fault and classify the fault type by the feature frequency. In addition, they proposed a method to perform signal processing by decomposing the envelope EMD to improve the diagnostic accuracy. However, this method is computationally intensive due to the iterative computation based on the reduced averaging method, and the convergence rate is different depending on the signal characteristics, which makes it difficult to perform real-time functions and consume a lot of memory space for data communication. In this paper, a method for diagnosing faults in rolling bearings based on compressive sensing (CS) and local characteristic-scale decomposition (LCD) is proposed and the effectiveness of bearing fault diagnosis method is verified by numerical experiments. In this paper, we propose a method to improve the diagnostic accuracy and shorten the computational time by identifying characteristic frequencies of the bearing fault from the Hilbert envelope spectrum of the components decomposed by the LCD after preprocessing and signal filtering of vibration signals based on CS.