Rapid Analysis of ICUMSA Value of Cane Sugar Using Multi-Channel Spectra Sensor Based-Portable Device

Abstract

One of important quality parameters of white crystal cane sugar is its color measured as ICUMSA (International Commission for Uniform Methods of Sugar Analysis) color value. It is usually measured in laboratory through a complex and lengthy chemical analysis method. To overcome this challenge, this research tries to explore the potential use of multi-channel spectra sensors in UV-VIS-NIR regions as an alternative method to predict the ICUMSA value. The proposed portable device uses a AS7265X sensor as its main component. After that, measurements were made in the laboratory using standard methods as reference data. The result of the prediction with partial least squares (PLS) is R² = 0,896, RMSEC = 0,072%, RMSEP = 0,103%, CV=26,09% and PRD = 3,10. Multiple linear regression (MLR) predictions are R²= 0,935, RMSEC=0,057%, RMSEP=0,090%, CV=20,64% and RPD = 3,92. The prediction of ICUMSA with an artificial neural network model (ANN) is R²=0,9996, RMSEC=0,004%, RMSEP=0,037%, CV=1,43% and the RPD value is 9,54. This shows that developed PLS, MLR and ANN are able to predict the ICUMA value, with ANN as the best model.