Fault diagnosis technique of induction machines with ordered harmonic and noise cancellation

In this paper, a fault diagnosis technique through fast Fourier transform (FFT) with suppressed the noise content is studied in greater detail. Because the line current utility driven motors involve interference components such as low order harmonics, high frequency disturbances, and the noise generated by harsh industrial environment, the fault analyses yield erroneous or fluctuating fault signatures. Mostly, the amplitudes of these components are higher than those of the fault signature itself and neither their existence nor behavior is predictable. The precision of the fault diagnosis methods mostly suffers from these high energy components and the ambiguity caused by them. Here, a new method is proposed to monitor and detect the fault signature through the noise ambiguity estimation of a motor current signal, which is cooperatively used to further enhance the fault diagnosis capability and reliability of the familiar FFT technique. The proposed algorithm is mathematically derived in detail and experimentally verified using a 3-hp motor generator setup.