Diagnosis of induction machines under non-stationary conditions by means of the spectral filter

Abstract

This paper introduces a new way to extract the fault components of a signal by means of a new proposed filter based on the Sampling Theorem and the properties of Discrete Fourier Transform. The proposed method is able to extract fault components either in steady-state or in transient conditions. This paper also proposes a new way to compute the energy of a signal in the frequency domain by means of the Plancherel's Theorem. Unlike the Discrete Wavelet Transform, which makes the filtering process in the time domain, the Discrete Fourier Transform allows the filtering process in the frequency domain. The use of the Discrete Fourier Transform makes the filtering process possible without applying any pretreatment to the current as it is done in the former works. As a consequence of the filtering process in the frequency domain and in order to reduce useless computations, in the current paper is also demonstrated the possibility of the computation of the energy of the extracted fault component in the frequency domain. The results are compared with the former and already published works in the area showing the good reliability of the new filtering process and energy computation in the frequency domain.