摘要：Recursive state estimation methods have aroused substantial attraction among many researchers and in particular, the drives research fraternity has shown increased interest in recent years. State estimators that surrogate direct measurements play an integral part in the operation of modern a.c. drives. Their robustness and accuracy are very much decisive for the performance of the drive. In this paper, a comparative analysis of the three nonlinear filtering schemes to estimate the states of a three phase induction motor on the simulated model is presented. The efficacy of Ensemble Kalman Filter (EnKF) against the traditional Jacobian based Filter or Extended Kalman Filter (EKF) and almost forbidden, hitherto least-attempted Unscented Kalman Filter (UKF) is very much exemplified. Theoretical aspects and comparative simulation results are investigated comprehensively with respect to three different scenarios viz., step changes in load torque, speed reversal, and low speed operation. Also, “Monte Carlo Simulation” runs have been exploited very extensively to show the superior practical usefulness of EnKF, by which the minimum mean square error (MMSE), which is often used as the performance index, ostensibly gets mitigated very radically by the proposed approach. The results throw light on alleviating the intrinsic intricacies encountered in EKF in parlance with the observer theory.