Pendugaan Kadar Lemak dan Kadar Air Biji Kakao Aksesi Papua Barat Menggunakan Near Infrared Spectroscopy (NIRS)

Abstract

Kadar lemak dan kadar air pada biji kakao sangat penting dalam perdagangan karena menjadi tolok ukur penentuan harga biji kakao. Namun demikian pengukuran kadar lemak dan kadar air biji kakao masih dilakukan secara destruktif sehingga kurang efisien. Dengan kemajuan teknologi, pendugaan kadar lemak dan kadar air biji kakao dapat dilakukan secara non-destruktif menggunakan teknologi NIRS. Tujuan penelitian ini yaitu mengembangkan metode pendugaan kadar lemak dan kadar air biji kakao secara non-destruktif menggunakan teknologi NIRS. Prosedur penelitian meliputi persiapan sampel, pengukuran spektrum biji utuh, penggilingan biji kakao, pengukuran spektrum bubuk, analisis proksimat, pre-treatment dan pengolahan spektrum, serta kalibrasi dan validasi model pendugaan kadar lemak dan kadar air biji kakao. Spektrum dianalisis menggunakan Partial Least Squares Regression. Pre-treatment spektrum yang digunakan yaitu Smoothing Savitzky-Golay, Multiplicative Scatter Correction, Standard Normal Variate, Normalization, Baseline Offset, dan De-trending. Perlakuan terbaik dalam menduga kadar lemak diperoleh dari bentuk bubuk menggunakan spektrum reflektan dengan pre-treatment SSg, menghasilkan nilai SEC 2,92%; 2,66%; CV 6,62%; RPD 1,51; konsistensi 109,96%; dan r 0,72. Perlakuan terbaik dalam menduga kadar air diperoleh dari bentuk biji utuh menggunakan spektrum reflektan dengan pre-treatment BO, menghasilkan nilai SEC 0,35%; SEP 0,33%; CV 5,64%; RPD 1,53; konsistensi 105,89%; dan r 0,81.

Fat content and water content are very important in trading because they determine the price of cocoa beans. However, fat and water content measurements are still conducted destructively, making it less efficient. With the advancement of technology, prediction of fat and water content can be done non-destructively using NIRS technology. The objective of this research is to develop a non-destructive method for predicting the cacao beans fat and water content using NIRS technology. Research procedures included sample preparation, whole bean spectrum measurement, cacao beans grinding, powder spectrum measurement, proximate analysis, pre-treatment and spectrum processing, and calibration and validation model for predicting cacao beans fat and water content. Spectrum was analyzed using Partial Least Squares Regression. Pre-treatment used was Smoothing Savitzky-Golay, Multiplicative Scatter Correction, Standard Normal Variate, Normalization, Baseline Offset, and De-trending. The best method for determining fat content is using reflectance spectrum in powder form with SSg pre-treatment. produced SEC 2.92%; SEP 2.66%; CV 6.62%; RPD 1.51; consistency 109.96%; and r 0.72. The best method for determining water content is using reflectance spectrum with BO pre-treatment to estimate water content in whole bean form, produced SEC 0.35%; SEP 0.33%; CV 5.64%; RPD 1.53; consistency 105.89%; and r 0.81.