Non-destructive measurement of sugar content in apples using millimeter wave reflectometry and artificial neural networks for calibration

A millimeter wave reflectometer has been developed for non-destructive measurement of sugar content in fruits. The intensity of reflected wave varies not only by sugar content but also by temperature of fruits. Therefore, a calibration curve to calculate sugar content by binary measurements is necessary. We report here a method for making a calibration curve by using a feed forward artificial neural network (ANN). An ANN containing three layers of nodes was trained. Sigmoidal and linear transfer functions were used in the hidden and output layers, respectively. The proposed method was applied satisfactorily to measure sugar content in apples.