In this paper, first we refom1Ulated the finite element model
(FEM) into a neural network structure using a simple two - dimensional problem. The structure of this neural network is described
, followed by its application to solving the forward and inverse problems. This model is then extended to the general case and the advantages and di sadvantages of this approach are descri bed along with an analysis of the sensi tivity of the algorithm to errors in the measurements. Consider a typical boundary value problem with the
govern ing differential equation: Lcp = f, where L is a differential
operator, f is the forcing function and cp is the unknown quant ity. This
di fferential equation can be solved in conjunction wi th boundary conditi ons on the boundary r enclosing the domain. A commonly
used approach to solve this problem is to use the finite element approach.
