Fault diagnosis and condition monitoring of axial-flux permanent magnet wind generators

Abstract

This paper presents fault detection techniques in an attempt to discriminate between two common faults in axial-flux permanent magnet wind generators namely; static eccentricities and interturn short circuit faults. A combination of condition monitoring systems is proposed to achieve the discrimination; with detection techniques derived from current and vibration signatures for both faults using signal processing techniques. Parametric spectral estimation technique is applied as an alternative to the Fourier transform to process the fault signatures. This effectively overcame the problem of non-stationarity of signals due to variable wind speed as high frequency resolution was attained with less than three seconds of measurement data. The results will advance the development of standards and technical guidelines in the condition monitoring of permanent magnet synchronous generators in wind turbines.