Parameter estimation of nonlinear discrete-time dynamic systems using genetic algorithm

Abstract

The main important thing about modelling a system is to understand the behaviour and to aid in designing a system (Soderstrom and Stoica, 1989). There are several steps to be considered in constructing and identifying a system's behaviour (Ljung, 1999). The first step is data acquisition which is to gather input and output data of the system to be identified either from experiment data through real process plant or simulated data. Then, a model structure consists of mathematical equation that describes the behaviour of a system is chosen to represent the system. The next step is to estimate the parameters of the model structure to complete the mathematical representation of the system. Finally, model validation is done to verify whether the identified model structure fulfils the modelling requirement according to objective criteria of a good model approximation.