Classification of Plastic Materials using a Microwave Negative-Order-Resonance Sensor and Support-Vector-Machine

Abstract

A method for plastic material classification using a negative-order-resonance (NOR) sensor operating at the 2.5 GHz band and support-vector-machine (SVM) for pattern recognition is presented. The proposal experimentally demonstrates the correct classification of different plastic materials based on their dielectric properties, dealing with large sources of uncertainty introduced by pellet measurements such as air gaps and position/dimension of the pellets. The proposed technique results attractive for the plastic industry as it involves a fast and nondestructive process along with the use of small circuit elements.