Size Parameters Classification of Loop Current Model of PCB Traces by CNN of Magnetic Near-Field Distribution

Abstract

A method of an equivalent loop current modeling of printed circuit board traces and components on printed circuit boards was investigated. A microstrip line (MSL) and parallel lines are used as test benches. The loop current is one of common electromagnetic field source models, and is composed of a signal and a return current in this case. One-dimensional convolutional neural network (CNN) was used as a classifier to estimate the size parameters of the loop current model. The results show a feasibility and necessity for improvement of the equivalent loop current modeling by the machine learning of the near-field information.