Identification of Dynamical Systems Through the Structure of Auto-regression With Exogenous Variable by Decreasing Gradient and Least Squares

Abstract

In this article was made the identification of dynamic systems of first and second order more common in electronics such as low and high pass filters of the first order, pass-band filter and direct current motor through the structure of auto-regression with exogenous variable. The proposed dynamical systems are initially modeled by a continuous-time transfer function using physical laws. Subsequently, a step entry signal was applied and the data for the identification process was recorded in discrete time. The estimation of parameters was performed with the method of decreasing gradient and least squares. It was obtained as a result that the least squares method could not find a model for the first-order high-pass filter, but the decreasing grade method allowed to model all the proposed systems.