CELLULAR NEURAL NETWORKS FOR EDDY-CURRENT PROBLEMS

Up to now the finite element analysis (FEA) run on digital computers has been the most common tool for solving eddy current problems [3, 6-10]. However, this solution suffers from two serious drawbacks. First, due to the serial processing nature of the digital computers, analysis is often time consuming. Second, the widespread access to the better developed commercial software is limited by its high cost. In this paper, we will present a new high speed, parallel processing alternative, with an inexpensive simulation option, based on cellular neural networks. This alternative has potential applications to a broad area of problems which can be described by partial differential or integro-differential equations. Examples are problems in heat transfer and electromagnetic fields.