Estimating the Heat Transfer Coefficient Using Universal Function Approximator Neural Network

Abstract

Abhstract-The appropriate knowledge of the Heat Transfer Coefficient (HTC) is required for the efficient design of heat treatment operations. There are several inverse heat transfer calculation methods to determine this quantity, but these are usually based on heuristics search algorithms and require high computation demands. This paper presents a solution to this problem with special usage of Artificial Neural Networks (ANN), the universal function approximator. After the time-consuming training process, this network is capable of giving prompt estimations about the nature of the HTC function sought. This estimation would be a useful input for additional fine-tuning algorithms.