Machine Condition Monitoring by Online Updated Optimized Weights Spectrum: An Industrial Motor Case Study

Abstract

Machine condition monitoring (MCM) is beneficial to gaining more profits and avoiding unexpected incidents, which has received much attention from the academic and industrial fields. Fault feature extraction is crucial for MCM. Recently, an optimized weights spectrum (OWS) is proposed to extract fault features in the Fourier spectrum, however, the calculation of the OWS is restricted by the usage of fault signals. This paper proposed an online updated OWS to relieve the usage of fault signals, and a 3D OWS can be obtained to exhibit the run-to-failure fault features in the Fourier spectrum. What's more, instead of man-made experimental run-to-failure datasets, a motor dataset collected from an industrial coal mining factory validated the performance of the online updated OWS for MCM.