Application of the support vector machines method for metal analyse by an eddy current system

Abstract

Abstract The eddy current method of control analyzed and the possibility of its use for the identification of metals and alloys substantiated. It is shown that this can be done by modernizing the eddy current radio engineering system with dynamic scanning of metal samples, which consists in the use of modern microcontrollers with high‐precision ADC operation. The structural of the modernized system is presented, some aspects of processing the received response signal are considered, which made it possible to highlight the information coefficients that would represent the metal or alloy as a generalized image. It is proposed to use machine learning methods to compare an unknown sample of a metal or alloy under study with those available in the collected database, which is shown on the example of the support vector machine method, which allows for real‐time identification of metals and alloys, to carry out express control, without human intervention in the identification process.

Keywords

• machine learning
• metals
• support vector machine
• system identification