Nondestructive Inspection of Melon's Sugar Content Based on Impedance Characteristics

摘要

The objective of the study is to determine melon sweetness by using the impedance characteristics. 34 samples are used as the calibration set to training and the rest 16 melons are used as validation set. First the equivalent series resistor and equivalent series capacitor of melons are measured over the frequency from 1 KHz to 100 KHz. Then the sweetness is measured by a portable refractometer. Through principal component analysis (PCA), principal components are selected to model by back propagation neural network (BPNN) optimized by genetic algorithm (GA). Comparing this method with BPNN and partial least squares (PLS), it is obviously showed that PCA-BPNN optimized by GA model is reliable and practicable. The inspecting results are assessed by correlation coefficient R = 0.944, and the root mean squares error of prediction RMSEP = 0.707. The method is proposed to detect melon's sweetness.