Identification of an Induction Motor with the Least-Squares Method

This paper introduces the theoretical analysis of the least-squares method applied to the identification of an induction motor directly supplied by the grid but valid for any supply condition. It is shown that the estimation of the electrical parameters of the motor can be performed by unconstrained minimisation, indirectly taking into consideration the constraints which inevitably arise when the well-known stator and rotor voltage equations are rearranged so as to allow the application of the least-squares method. In this way no convergence proof is necessary. The assumptions under which this technique is valid are also presented as well as the identifiability criteria both for transient and sinusoidal steady-state. Moreover some practical issues about the choice of the most suitable reference frame are also developed on the basis of numerical considerations.