Applying neural networks to the solution of forward and inverse heat conduction problems (Q865761)

Approximating the space derivative, one and two dimension heat conduction models are reduced to Hopfield differential equations. Numerical solutions to the forward heat conduction problems are then obtained using the continuous Hopfield neural network approach. The obtained results are used as training data for a three layered MLP trained with different types of back propagation algorithms. Further, as well known, this network will be able to make predictions and it will be designed to solve inverse heat conduction problems. Different numerical simulations are shown.