Modeling of a Flexible Beam System using NARMAX Model integrated with Multi-Objective optimization Differential Evolution

Abstract

Modeling a dynamic system such as a flexible beam system without knowing the physical equations involved is a challenging task. An accurate model representing the system is needed for more efficient controller design. This paper presents the identification of a flexible beam system using Nonlinear Auto- Regressive Moving Average with eXogenous input (NARMAX) model integrating with Multi-Objective optimization Differential Evolution (MOODE) algorithm. This proposed algorithm called MOODE-NARMAX is applied to minimize two objective functions; the model complexity and the mean square error between actual and predicted outputs. The system identification procedures were followed for identifying the flexible beam system. Model validity tests were applied to the set of solutions called the Pareto-optimal set that was generated from the proposed algorithm to select an optimal model. Using the input and output experimental data of the rig, the simulation results show that the MOODE-NARMAX algorithm is able to produce an adequate and optimal model for the flexible beam system and therefore could be used as an alternative to identify an adequate and parsimonious model for the system.