The Detection and Quantification of Adulteration in Ground Roasted Asian Palm Civet Coffee Using Near-Infrared Spectroscopy in Tandem with Chemometrics

2nd International Conference on Agricultural Engineering for Sustainable Agricultural Production

Suhandy, D.^1 ; Yulia, M.^2 ; Ogawa, Y.^3 ; Kondo, N.^3

Department of Agricultural Engineering, University of Lampung, Indonesia^1

Department of Agricultural Technology, Lampung State Polytechnic, Indonesia^2

Laboratory of Bio-Sensing Engineering, Kyoto University, Japan^3

关键词: Calibration samples;    Cross-validation methods;    Detection and quantifications;    Determination coefficients;    Diffuse reflectance measurements;    Integrating spheres;    Partial least square (PLS);    Root-mean-square error of predictions;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/147/1/012011/pdf
DOI  :  10.1088/1755-1315/147/1/012011

来源: IOP

【 摘 要 】

In the present research, an evaluation of using near infrared (NIR) spectroscopy in tandem with full spectrum partial least squares (FS-PLS) regression for quantification of degree of adulteration in civet coffee was conducted. A number of 126 ground roasted coffee samples with degree of adulteration 0-51% were prepared. Spectral data were acquired using a NIR spectrometer equipped with an integrating sphere for diffuse reflectance measurement in the range of 1300-2500 nm. The samples were divided into two groups calibration sample set (84 samples) and prediction sample set (42 samples). The calibration model was developed on original spectra using FS-PLS regression with full-cross validation method. The calibration model exhibited the determination coefficient R2=0.96 for calibration and R2=0.92 for validation. The prediction resulted in low root mean square error of prediction (RMSEP) (4.67%) and high ratio prediction to deviation (RPD) (3.75). In conclusion, the degree of adulteration in civet coffee have been quantified successfully by using NIR spectroscopy and FS-PLS regression in a non-destructive, economical, precise, and highly sensitive method, which uses very simple sample preparation.