A new method for identifying diagnostic rich frequency bands under varying operating conditions

Performing condition monitoring under time-varying operating conditions is challenging. The varying operating conditions impede the ability of conventional fault diagnosis methods to detect damage on rotating machine components such as bearings and gears. This paper investigates a new method for identifying diagnostic rich frequency bands under time-varying operating conditions. This method uses the orderfrequency spectral coherence and a feature, which is dependent on the cyclic order of interest and the frequency resolution of the spectral coherence, to decompose the signal into a feature plane. Thereafter, the spectral frequency and the spectral frequency resolution that maximise the feature plane are used to design a bandpass filter. The bandpass filter extracts a diagnostic rich signal, which can be analysed by using the squared envelope spectrum or the synchronous average. The proposed method is compared to the fast kurtogram on a numerical gearbox dataset as well as on an experimental gearbox dataset, with very promising results obtained.

Jaar van publicatie:2019

Toegankelijkheid:Open