A neural network inversion approach to electromagnetic device design

Abstract

In this paper we present a new model that employs, in a natural and effective way, a neural network inversion algorithm providing a solution of the electromagnetic device design problem. The model combines two types of artificial neural networks and applies a neural network inversion algorithm. Further development of this model can propose an efficient solution to the electromagnetic device design problem. The model is applied to the magnetic stimulation coil design as well as gradient coil design. The results obtained show the effectiveness of the proposed model.