The increase in size of the air cell is related to the aging process of the eggs. According to the European Commission Regulation, eggs must be classified in A (air cell size higher than 4 mm), and A “extra” (air cell size lower than 4 mm) categories by candling inspection. This technique is unable to non-destructively assess the size of the air cell during egg grading. The present research studies the possibility to nondestructively grading shell eggs from dielectric parameters obtained by means of a sine wave RF oscillator, a parallel inductance and capacitance circuit. In particular, dielectric parameters and egg dimensional characteristics were used to set up multi-layer (MLP) artificial neural networks. Using MLP with two hidden layers, eggs can be correctly graded in A and A “extra” categories (test validation) within a mean performance close to 90%.
artificial neural networks, dielectric properties, eggs.