A Comparison Between Independent Component Analysis and Established Signal Processing Methods for Gearbox Fault Diagnosis Under Time-Varying Operating Conditions

Abstract

Reliable condition monitoring methods are required for rotating machines operating under time-varying operating conditions. The measured vibration signals typically contain information related to the different interacting components (e.g. gear mesh components, bearing fault components), the transmission paths between the excitation sources and the sensors, the environmental conditions (e.g. changes in temperature) and the operating conditions of the machine. Hence, multiple sources could be present in the measured signals, which could impede the detection of weak sources attributed to incipient damage. Several methods have been proposed to solve this problem, including, synchronous statistics (e.g. time-synchronous averages, synchronous average of the squared envelope, synchronous median of the squared envelope), the squared envelope spectrum, the order-frequency spectral coherence and the integrated squared spectral coherence (e.g. the enhanced envelope spectrum and the improved envelope spectrum). Independent Component Analysis (ICA) is a well-established technique that has not been compared against the aforementioned methods. In this work, we compare the performance of ICA against the performance against established signal analysis methods for fault detection under time-varying operating conditions. We show that ICA performs well against established signal analysis-based condition monitoring methods for machines operating under time-varying conditions.