Neural network integrated physics driven deep learning model for the thermal response analysis of the pyramidal fin with internal heat generation

Pyramidal fins are widely used in heat sinks to enhance heat dissipation from electronic equipment, benefiting from their tapered design, low mass, and effective thermal performance. In view of this, the present study examines the impact of convection on unsteady heat transfer in both pyramidal and rectangular fins. Additionally, the internal heat generation and temperature-dependent thermal conductivity are considered for both cases. The dimensional time-dependent temperature equations of both fins are converted to a non-dimensional form with the aid of appropriate dimensionless terms. The physics-informed neural network (PINN) is utilized to evaluate the heat transport properties for various parameters. By integrating physical conditions and data learning, PINN improves thermal predictions. The artificial intelligence (AI) based design exhibits excellent agreement with finite difference method (FDM) outcomes and displays relatively low errors. The influence of various parameters on the thermal profile is visually illustrated for both pyramidal and non-pyramidal (rectangular) fin cases. A 17.64 % increase in the convective-conductive parameter results in a 21.49 % increase in the heat transfer rate.