System identification of essential oil extraction system using Non-Linear Autoregressive Model with Exogenous Inputs (NARX)

Abstract

This paper explores the application of Non-Linear Autoregressive Model with Exogeneous Inputs (NARX) system identification of an essential oil extraction system. Model structure selection was performed using the Binary Particle Swarm Optimization (BPSO) algorithm by (J. Kennedy and R. Eberhart, 1997). The application of BPSO for model structure selection represents each particle's position as binary values. Then, the binary values were used to select a set of regressors columns from the regressor matrix. QR factorization was used to estimate the parameters of the reduced regressor matrix. Tests performed on the essential oil extraction system by (Rahiman, 2009), defined the 2nd order model with three terms, while fulfilling all model validation criterions.