On the identifiability of steady-state induction machine models using external measurements

Abstract

A common practice in induction machine parameter identification techniques is to use external measurements of voltage, current, speed, and/or torque. Using this approach, it has been shown that it is possible to obtain an infinite number of mathematical solutions representing the machine parameters. This paper examines the identifiability of two commonly used induction machine models, namely the T-model (the conventional per phase equivalent circuit) and the inverse Γ-model. A novel approach based on the alternating conditional expectation (ACE) algorithm is employed here for the first time to study the identifiability of the two induction machine models. The results obtained from the proposed ACE algorithm show that the parameters of the commonly employed T-model are unidentifiable, unlike the parameters of the inverse Γ-model which are uniquely identifiable from external measurements. The identifiability analysis results are experimentally verified using the measured operating characteristics of a 1.1-kW three-phase induction machine in conjunction with the Levenberg-Marquardt algorithm, which is developed and applied here for this purpose.