An evolutionary intelligent approach for the LTI systems identification in continuous time

Abstract

Identification and modeling of systems are the first stage for development and design of controllers. For this purpose, as an alternative to conventional modeling approaches we propose using two methods of evolutionary computing: Genetic Algorithms (GA) and Particle Swarm Optimization (PSO to create an algorithm for modeling Linear Time Invariant (LTI) systems of different types. Integral Square Error (ISE) is the objective function to minimize, which is calculated between the outputs of the real system and the model. Unlike other works, the algorithms make a search of the most approximate model based on four of the most common ones found in industrial processes: systems of first order, first order plus time delay, second order and inverse response. The estimated models by our algorithms are compared with the obtained by other analytical and heuristic methods, in order to validate the results of our approach.