Harnessing Deep Learning to Solve Inverse Transient Heat Transfer With Periodic Boundary Condition

Abstract

Accessing temperature data in certain manufacturing and heat treatment processes can be a challenge. Inverse heat conduction problems (IHCPs) offer a solution, allowing us to determine temperatures in inaccessible locations using transient temperature or heat flux measurements from accessible surfaces. This study investigates the capability of a deep neural network (DNN) approach for predicting the front surface temperature and heat flux from the measured back surface temperature and heat flux. The back surface temperature and heat flux are determined using a direct python script code. The inverse solution is then applied with the help of the fully dense DNN approach. To prevent overfit and nongeneralization issues, the regularization and dropout techniques are embedded into the neural network framework. The results reveal that the DNN approach provides more accurate prediction compared to the previous mathematical frameworks such as the conjugate gradient method (CGM). Moreover, the model is tested by noisy data (from 1% to 10%) causing instabilities in the recovered front surface conditions. Despite the presence of noise, the model can overcome this difficulty and is able to predict the desired parameters with a good accordance. Another significant potential of the developed model is its unique capability to deal with the highly periodic heat flux at boundary conditions.