Unbalanced transients-based maximum likelihood identification of induction machine parameters

The paper describes an effective formulation of a maximum-likelihood identification algorithm for linear estimation of the equivalent-circuit parameters of cage-type (single cage and double cage) or deep-bar induction motors with measurement and process noises. A complete generalized model for symmetrical and asymmetrical test analysis of induction machines is developed for this purpose. The paper outlines the theory and reasoning behind the proposed statistical-based treatment of online data derived from generalized least-squares estimator and a Kalman filter. The method is successfully applied to online double-line independent finite-element (FE) short-circuit-simulated records of a deep-bar-type induction motor.