Department of Mechanical and Biosystem Engineering, IPB University
Sutrisno MARJAN
Department of Mechanical and Biosystem Engineering, IPB University
Nissa ADIARIFIA
Department of Mechanical and Biosystem Engineering, IPB University
Inna NOVIANTY
Study Program of Computer Engineering, Vocational School, IPB University
Yunisa Tri SUCI
Department of Mechanical and Biosystem Engineering, IPB University
(*) Corresponding authors:
wbudiastra@apps.ipb.ac.id |
I Wayan BUDIASTRA
Abstract
The study used near-infrared reflectance (NIR) spectroscopy and hybrid calibration methods to predict oil and free fatty acid (FFA) content of Oil Palm Fruitlets (OPF) non-destructively. The reflectance and chemical content of OPF were measured and the calibration between NIR spectra and chemical content was performed using hybrid calibration methods (PLS-ANN, PCA-ANN). The best models to predict oil and FFA content of OPF respectively were the hybrid model of PLS-ANN with 25 Factor Components (FC) (R2 = 0.96; SEP= 2.21%, RPD = 4.79) and 19 FC (R2 = 0.96; SEP= 0.25%, RPD = 4.24) using Savitzky-Golay first derivative spectra pre-treatment.
Abstract in Indonesian
Penelitian ini menggunakan spektroskopi inframerah dekat (NIR) dan metode kalibrasi hibrid untuk memprediksi kandungan minyak dan asam lemak bebas buah sawit secara nondestruktif. Reflektan dan kandungan kimia buah sawit diukur dan kalibrasi antara spektra NIR dan kandungan kimia dilakukan menggunakan metode kalibrasi hibrid (PLS-ANN, PCA-ANN). Model terbaik untuk memprediksi masing masing kandungan minyak dan ALB buah sawit adalah model hibrid PLS-ANN dengan input 25 Factor Component (FC) (R2 = 0,96; SEP = 2,21%, RPD = 4,79) dan 19 FC (R2 = 0,96, SEP = 0,25%, RPD = 4,24) menggunakan pretreatment spektra turunan pertama Savitzky-Golay