Font Size: 

Measurement of total sugar and moisture content of granulated coconut sugar are usually carried out destructively by chemical analysis and gravimetric methods. This study aimed to test the performance of a portable spectrometer based on the AS7265X optical sensor which is able to capture the reflectance of the Vis-NIR light spectrum. The granulated coconut sugar samples used in this study were produced from six treatments by adding cane sugar during coconut sap heating, i.e. 0%, 1%, 2%, 3%, 4%, and 5% (w/v). The Vis-NIR light reflectance of the produced coconut sugar was then measured using the developed portable spectrometer. For validation, the total sugar and water content samples of granulated coconut sugar were also measured destructively. The estimation of total sugar content and water content of granulated coconut sugar was carried out using the backpropagation neural network method. The results showed that the developed artificial neural network with 2 hidden layers can be used to estimate the total sugar and water content of granulated coconut sugar better than the artificial neural network with 1 hidden layer. The mean absolute percentage error (MAPE) of the developed artificial neural network with 2 hidden layers for the estimation of total sugar and water content were 0.62% and 12.89%, respectively.