Assessment of strategies and potential for neural networks in the inverse heat conduction problem

Abstract

The inverse heat conduction problem (IHCP) has been identified as a good candidate for solution using artificial neural networks (ANNs) [5, 9]. Reasons that have been cited include the ability of ANNs to represent complex mappings of input/output vectors and the availability of a virtually unlimited amount of training data for the IHCP from solutions of the corresponding forward problem. This paper presents a synopsis of efforts to achieve successful solution of the IHCP via ANNs. Two basic approaches are presented: whole domain (whole history mapping) and sequential (local-time mapping). The experiences recounted here show some promise for the approach to the solution of inverse problems using ANNs and will be useful to others interested in application of these methods to this class of problems.