Electronic Theses and Dissertation

penentuan parameter kualitas internal yang cepat dan simultan seperti kadar air dan kadar lemak perlu diprediksi dalam pengolahan produk kakao. penelitian ini bertujuan untuk mengembangkan aplikasi teknologi near infrared spectroscopy (nirs) dalam memprediksi kualitas biji kakao utuh. data spektral inframerah dekat diakuisisi menggunakan instrument nirs portabel (psd nirs i16) dalam rentang panjang gelombang 1000-2500 nm. pengukuran kadar air menggunakan metode tidak langsung sedangkan pengukuran kadar lemak menggunakan metode soxhlet. metode koreksi menggunakan metode penghalusan spektrum standard normal variete (snv) dan de-trending orde 2 (dt-2). model regresi dibangun dengan partial least square regression (plsr) dan pricipal component regression (pcr). kandungan kadar air dan kadar lemak dapat di prediksi secara non-destriktif dan simultan tanpa melibatkan bahan kimia. penggunaan model snv+pcr dan snv+pls adalah metode terbaik pada penelitian ini dikarenakan memiliki hasil yang cukup baik untuk prediksi senyawa kimiawi pada biji kakao dengan nilai rpd kadar air 2,16 sedangkan kadar lemak 2,39.

Determination parameter quality internal Which fast And simultaneous like rate water And rate fat needs to be predicted in the processing of cocoa products. This research aims to develop technology applications near infrared spectroscopy (NIRS) in predicting the quality of whole cocoa beans. Near infrared spectral data were acquired using a portable NIRS instrument (PSD NIRS i16) in the wavelength range of 1000-2500 nm. Measurement of water content using the indirect method while measuring the fat content using the Soxhlet method. The correction method uses standard normal variete (SNV) and de-trending order 2 (Dt-2) spectral smoothing methods . The regression model was built using partial least square regression (PLSR) and principal component regression (PCR). The moisture content and fat content can be predicted non-destrictively and simultaneously without involving chemicals. The use of the SNV+PCR and SNV+PLS models is the best method in this study because it has good results for the prediction of chemical compounds in cocoa beans with an RPD value of water content 2.16 while the fat content is 2.39.