Estimation of Weight Fractions of Coffee Premix Ingredients using Taste Sensors

Abstract

　Incorrect ingredient weight fraction is an important concern of soluble coffee premix products. It might cause rejection and return of products. For this study, taste sensors were applied to estimate the weight fractions of ingredients in model soluble coffee premixes consisting of soluble coffee, cream powder, and granular sugar, aimed at application for inspecting returned products. Nine tastes (acidity, bitterness with rough taste, astringency, umami, saltiness, bitterness, astringency with stimulating taste, umami richness, and sweetness) were evaluated using six sensors using the change in membrane potential caused by adsorption (CPA) measurement method. Linear regression analysis using data of single taste yielded good correlation only with the soluble coffee weight fraction. Using all nine tastes in combination with partial least squares (PLS) regression analysis, the weight fractions of all ingredients (soluble coffee, cream powder, and granular sugar) were estimated with reasonable accuracy.