Identificación de sistemas en lazo cerrado basada en una estrategia híbrida AGA-Simplex

Abstract

Closed-loop identification of continuous systems, which can be considered as a nonlinear optimization problem, may result in a difficult solution problem when conventional methods are used. In this paper it is presented a hybrid strategy based on an Adaptive Genetic Algorithm and the Simplex method, that results in a satisfactory solution for this problem. The proposal is compared with other techniques reported in the literature. Three examples show the performance of the method: identification of high order dynamics; identification of unstable second order dynamics in open-loop; and parameter estimation in power generation systems. Simulation results show that the proposed is a robust method for close-loop system identification.