Wind turbine condition monitoring by detecting the transient period of a single generator stator current signal

Abstract

A wind farm often consists of a number of wind turbines (WTs) operating simultaneously, which not only challenges the operation of the wind farm but makes the condition monitoring (CM) more costly than ever before if each turbine is equipped with a standard commercially available WT CM system (e.g. SKF's WindCon system). With the continual increase of wind farm size, the operator is under increasing pressure to lower the CM cost so that the energy cost can be accordingly reduced in the end. The research presented in this paper is for meeting such a requirement by developing a new cost-effective CM technique based on detecting the transient period of a single WT generator stator current signal. Instead of collecting multiple numbers of CM signals from WT key components, the new technique relies on the use of only a single electric current signal collected from WT generator stator. Moreover, the new technique directly deals with the time waveform of the signal without using any artificial transforms (e.g. Fourier transform). Therefore, it is potentially able to provide more reliable CM result. Experiments have shown that the proposed technique is valid for detecting both the electrical and mechanical faults emulated on a specially designed WT drive train test rig.