Mechanical Fault Recognition for WTG based on Aligning Vibration Signal's Energy Spectrum

Fault recognition for Wind Turbine Generator's (WTG) mechanical system is a fundamental task of Condition Monitoring System and depends on the time-frequency feature of the monitored signals. A recognition method by matching the Hilbert-Huang Spectrum (HHS) of the vibration signal is proposed. Fault recognition is formulated as an aligning process that shifts the vibration's HHS to match the pre-defined fault pattern using ICP. Given that the duration and initial frequency feature of the HHS may lead to different aligning results, iteration parameters with multiple time-scales and positions are set. In simulations, a mechanical fault is identified and the effectiveness is compared.