Validation of the Effectiveness of Near-Infrared Spectroscopy for Detecting Impurities in Milk Powder using ANN and SVM

Abstract

Nowadays, there are so many productions on the market that based on dairy product that include milk powder. The milk powder has different categories which is for children, adult and others. The milk powder involves the local product until the import product. The transparency of the content in the milk powder cannot be identified easily and can cause food poisoning to be happened. Hence, with the use of near-infrared spectroscopy, NIRS it can be use easily to detect the content in the food simply. There are many methods in NIRS which is reflectance and absorbance. This method being uses to detect which milk powder is adulterated. Ocean Optic XR-1 ES, USA device and Ocean View software have been used in this project to take data from the sample. The device use wavelength between 200nm-1025nm. The milk powder obtained from the supermarket and nearby stores. Data that have been taken then transferred into suitable table and will go through two different analysis method which is SVM and ANN. The reflection method is not suitable to detect the adulterated milk powder. From the result ANN shows the higher accuracy to classify the pure and Adulterant milk powder.