Algebraic identification of a DC servomechanism using a Least Squares algorithm

Abstract

This paper proposes a parameter identification methodology based on a discrete-time Least Squares algorithm and a parametrization obtained using the Operational Calculus. References [4] and [5] proposed previously this parametrization and developed an Algebraic Identification Method (AIM) for parameter estimation of linear systems. The AIM employs the Operational Calculus to obtain analytical expressions for the parameter estimates. These expressions have a singularity at t=0 and certain excitation signals may also produce singularities at other time instants. The proposed approach employs the same parametrization obtained using the Operational Calculus, which is linear in the parameters, and employs a standard on-line discrete-time Least Squares algorithm. In this way, the proposed approach completely eliminates the problem of singularities; moreover, it is experimentally shown that the AIM and the proposed approach have similar performances.