Predictioin of fruit quality using near-infrared VIA NARX model

Abstract

Nowadays, the awareness of health and safety among consumers has been increase. This scenario caused them become willing to pay more for high quality fruit products. However, it is not easy to grade fruits by using only our eyes. Therefore non-destructive fruits internal quality assessment technique is an area that both technology and market section concern about. The objectives of this project are to study about Near Infrared Spectroscopy (NIRS) as a fruit quality measurement method, to evaluate the use of NIRS for nondestructive measuring SSC of apples and to predict the best model of the measurement data by using Auto-Regressive with Exogenous Input (ARX) Model and Nonlinear Auto-Regressive with Exogenous Input (Nonlinear ARX) Model. The Near-Infrared (NIR) reflectance spectra and the soluble solids content (SSC) of apples data have been recorded before from an experiment. The impact of the orders or numbers of poles of the model has been investigated based on the performance (best fit) of the model. The ARX Model and Nonlinear ARX Model indicate excellent prediction performance of the model with the best fit value were 87.11% and 100% respectively.