An enhanced instantaneous angular speed estimation method by multi-harmonic time–frequency realignment for wind turbine gearbox fault diagnosis

Abstract

Gearboxes are one of the most critical transmission devices in wind turbines (WTs). Order tracking (OT) of variable-speed WT gear is a powerful means to carry out fault diagnosis. Since it is difficult to install a tachometer or encoder for gearboxes with limited structural space, it is impossible to obtain the instantaneous angular speed (IAS) of the gearbox. In recent years, many tacholess techniques have been presented to estimate instantaneous phase from vibration signals for OT. However, their performance needs to be further improved to get a more exact diagnosis result. To solve this issue, this paper presents an enhanced IAS estimation approach for WT tacholess order tracking (TLOT), which is available for accurate IAS estimation of the rotating shaft, so that TLOT can be more effectively conducted under conditions of large speed variation and background noise. An improved cost function algorithm is designed to extract the IAS from the vibration signal. A Vold–Kalman filter and Hilbert transform are combined to calculate the instantaneous phase. Afterwards, the raw non-stationary vibration signal is resampled at an equal angle increment, and the envelope order spectrum is calculated from the resampled signal for fault diagnosis of the WT gearbox. The main innovating idea of this approach is to improve the traditional cost function ridge-extraction method by introducing gradient information based on probability and multi-harmonic time–frequency realignment, which makes the IAS estimation process more accurate and overcomes the limitations of traditional IAS estimation techniques. The effectiveness of the presented approach is demonstrated by using a vibration signal measured from real-world WT gearboxes, and the validation results demonstrate that the presented approach surpasses some state-of-the-art methods in IAS estimation accuracy and efficiency.