Pendugaan Tingkat Fermentasi Kakao Secara Non-Destruktif dengan NIRS

Abstract

Deteksi tingkat fermentasi yang cepat dan efisien dapat diwujudkan melalui teknologi Near Infrared Spectroscopy (NIRS) dengan mengukur kadar gula reduksi pada biji kakao yang telah difermentasi. Tujuan penelitian ini adalah menentukan model kalibrasi terbaik dari metode Partial Least Squares (PLS) dan Principal Component Regression (PCR) yang dapat digunakan untuk menentukan kadar gula reduksi dan menganalisis korelasi kadar gula reduksi dengan lama fermentasi dari biji kakao. Korelasi data reflektan NIRS dengan kadar gula reduksi biji kakao diolah menggunakan lima pra pengolahan yaitu Normalization, Dg1, Dg2, Multiple Scatter Correction (MSC) dan Kombinasi Normalization dan Dg1. Metode PLS dengan pengolahan data MSC factor 7emiliki nilai R2, rk, SEC, rv, SEP, Konsistensi dan RPD masing-masing 0,7413, 0,9115, 0,0855, 0,9449, 0,1037, 82,4312, 2,1904. Metode PCR dengan pengolahan data Normalization faktor 9 memiliki nilai R2, rk, SEC, rv, SEP, Konsistensi dan RDP masing-masing 0,5823, 0,8511, 0,0002, 0,9222, 0,0002, 98,5492, 1,9550. Pendugaan kadar gula reduksi menggunakan NIRS dapat dilakukan. PLS lebih baik dari PCR untuk menduga kadar gula reduksi. Kadar gula reduksi dapat digunakan sebagai salah satu indikator berdasarkan analisis diskriminan. Namun, analisis diskriminan belum dapat mengkategorikan tingkat fermentasi berdasarkan hari fermentasi.

Fast and efficient detection of fermentation rates can be realized through Near Infrared Spectroscopy (NIRS) technology by measuring the reducing sugar content of fermented cocoa beans. The purpose of this study was to determine the best calibration model of the Partial Least Squares (PLS) and Principal Component Regression (PCR) methods that can be used to determine reducing sugar content and analyze the correlation of reducing sugar content with fermentation time of cocoa beans. The correlation of NIRS reflectance data with reducing sugar content of cocoa beans was processed using five pre-processing, namely Normalization, Dg1, Dg2, Multiple Scatter Correction (MSC) and Combination of Normalization and Dg1. The PLS method with MSC data processing factor 7 has the values of R2, rk, SEC, rv, SEP, Consistency and RPD respectively 0,7413, 0,9115, 0,0855, 0,9449, 0,1037, 82,4312, 2,1904. The PCR method with data processing Normalization factor 9 has R2, rk, SEC, rv, SEP, Consistency and RDP values of 0,5823, 0,8511, 0,0002, 0,9222, 0,0002, 98,5492, 1,9550. Estimation of reducing sugar content using NIRS can be done. PLS is better than PCR for predicting reducing sugar content. Reducing sugar content can be used as an indicator based on discriminant analysis with D = -2,589 + 5,654x, where x is reducing sugar content. Discriminant analysis has not been able to categorize the level of fermentation based on the day of fermentation.