Author :
Ferrah, A. ; Bradley, K.J. ; Asher, G.M. ; Woolfson, M.S.
Abstract :
Several techniques, analogue and digital, have been proposed for sensing the speed information from the motor terminal quantities, namely voltages and currents. Over the last three decades, many parametric methods of spectral estimation have been proposed to circumvent the inherent disadvantages of the FFT, i.e. lengthy data records, windowing, and interpolation. In this paper, in addition to the DFT techniques, four of these parametric methods are investigated with a focus on their ability to resolve the speed information carrier component and estimate its frequency. The techniques tested in this work are the FFT, the Goertzel algorithm, the maximum entropy method, and autoregressive adaptive techniques using the least-mean square algorithm, the fast recursive least-square algorithm, and the Kalman filter algorithm. Detailed mathematics were omitted for reasons of space. Computer simulations were carried out to assess and compare the performances of the methods. Then, a digitized stator current from a test rig was used to extend the comparison
Keywords :
computerised instrumentation; electric drives; harmonics; induction motors; least squares approximations; machine testing; rotors; velocity measurement; DFT; FFT; Goertzel algorithm; Kalman filter; autoregressive adaptive techniques; computerised instrumentation; digital simulation; fast recursive least-square algorithm; induction motor drives; least-mean square algorithm; machine testing; maximum entropy method; parametric methods; performances; rotor slot harmonics; spectral estimation; speed measurement;