ABSTRACT
This chapter provides a clear view of using the deep learning (DL) technique to establish a complete path to solve 3D passive/active steady heat conduction problems. It presents several recent works pertaining to using DL techniques to solve heat conduction problems. Traditional methods for solving the 3D steady heat conduction problem mainly include analytical methods and numerical methods. The variable separation method is one of the most used analytical methods for solving partial differential equations. For objects with irregular geometry shapes, numerical methods such as the finite difference method, the finite element method (FEM) and the finite volume method are widely used to solve the heat conduction equation. In terms of computing accuracy and efficiency, the proposed framework has exhibited apparent merits. The experimental results have firmly corroborated that the proposed model is able to give a precise prediction of the temperature field two orders of magnitude faster than traditional algorithms FEM.
