Monitoring of Induction Machines currents by high frequency resolution analysis

Diagnostic techniques of electrical faults for induction machines are usually based on spectral analysis of stator currents. Hence large time acquisition period are mandatory in order to achieve high frequency resolution. Moreover, a sharp knowledge of supply frequency and slip are required in order to monitor the spectrum lines specifically linked to electrical faults. Several research activities were aimed at retrieving algorithms to cope with the above issues. However for real-time implementation of diagnostic systems, a suitable computational cost is required. In this paper, statistical time domain techniques are used to track supply frequency and slip that features either a lower computational cost or a higher accuracy than traditional discrete Fourier transform techniques. Then, the accurate knowledge of supply frequency and slip are used to tune the parameters of the zoom-FFT algorithm that allows higher frequency resolution keeping constant the computational cost, or lower computational cost keeping constant the frequency resolution. Therefore, the proposed method allows an accurate estimation of the amplitudes of the stator current spectrum lines related to electrical faults at reduced computational cost