Abstract:
The order-frequency spectral coherence and its integrated spectra (e.g. improved envelope spectrum, squared envelope spectrum) are some of the most powerful methods for performing fault diagnosis under time-varying operating conditions. However, it may require much work to interrogate the order-frequency spectral coherence for symptoms of damage. Hence, in this work we propose a methodology that combines the order-frequency spectral coherence with historical data that were acquired from a healthy machine to obtain an anomalous envelope spectrum, which is further processed for fault diagnosis. This anomalous envelope spectrum is further processed with a smoothing operation to not only perform automatic fault detection, but it is also possible to identify the damaged component if the kinematics of the gearbox are known. The proposed method is investigated on one numerical gearbox dataset and three experimental datasets, where its potential for performing automatic fault detection under time-varying operating conditions is highlighted.
