A Novel Incipient Fault Detection Technique for Roller Bearing Using Deep Independent Component Analysis and Variational Modal Decomposition

Abstract Roller bearing failure can result in downtime or the entire outage of rotating machinery. As a result, a timely incipient bearing defect must be diagnosed to ensure optimal process operation. Modern condition monitoring necessitates the use of deep independent component analysis (DICA) to diagnose incipient bearing failure. This paper presents a deep independent component analysis method based on variational modal decomposition (VMD-ICA) to diagnose incipient bearing defect. On a newly established test setup for rotor bearings, fast Fourier techniques are used to extract the vibration responses of bearings that have been artificially damaged using electro-chemical machining. VMD techniques diminish the noise of the measurement data, to decompose data processed into multiple sub-datasets for extracting incipient defect characteristics. The simplicity of the VMD-ICA model enriched the precision of diagnosis correlated to the experimental results with weak fault characteristic signal and noise interference. Moreover, deep VMD-ICA has additionally demonstrated strong performance in comparison to experimental results and is useful for monitoring the condition of industrial machinery.