Speed estimation from noisy resolver signals

A Kalman filter based method for estimation of the rotor speed of an AC motor from sine/cosine resolver signals is presented. A very simple algorithm, requiring only six multiplications and two trigonometric operations, is obtained. The algorithm is easily implemented on a digital signal processor. Direct hardware implementation is also feasible. Performance evaluations using both simulated and experimental data are made. It is shown that no lagging of the estimated speed occurs at transients. The noise rejection is also adequate.