Wavelengths Selection Method in Near Infrared Spectra Using Multivariate Analysis for Nondestructive Determination of Oil Content in Palm Oil Fruits

Abstract

Fruit oil content (FOC) is one of the most significant commercial characteristics in oil palm output both in upstream and downstream phases. The NIR Spectroscopy approach was used as a method to determine the oil content of fresh oil palm fruits. Several tests on the NIR spectroscopy approach for estimating the oil content of oil palm fruits revealed that the accuracy was still ideal when utilizing earlier spectra processing. Still, some crucial information was lost when using this spectra processing. Five hundred samples, categorized into ten groups according to their maturity levels, were readied for reflectance measurements and chemical assessments of oil content. The NIRFlex N-500 FT-NIR Spectrometer was employed to measure the reflectance of the samples within the 1000-1500 nm wavelength range. The acquired spectrum of fresh oil palm fruits was then converted to absorbance (Log 1/R). The method applied in this research is to select NIR absorption wavelengths that correlate with palm fruit oil content using PCA analysis and then develop a model using MLR. Five influential wavelengths were identified for predicting oil content of oil palm fruit based on the highest PC values in the loading plot of PCA those are 1166.59 nm, 1188.2 nm, 1212.4 nm, 1387 nm, and 1486 nm. The wavelength of 1212 nm (Ar) was selected as the reference of absorbance for establishment of MLR model, as A1, A2 and A3 were absorbances at the wavelengths of 1251.88, 1252.51 and 1468.86 nm, respectively. The established MLR model of Y= -14088.57 (A1/Ar) + 14017.9 (A2/Ar) - 12.24 (A3/Ar) + 120.67 can predict oil content of oil palm fruit (Y) accurately (R2 of 0.8 and SEP of 3.28 %). These findings show that the regression model has a strong capacity for prediction of oil content of palm fruits easily and nondestructively without the need for spectral preprocessing, solvents, or reagents, making it environmentally friendly.