This repository presents a project that employs Physics-Informed Neural Networks (PINNs) to address challenging inverse problems in heat transfer. The project focuses on prototype heat transfer problems that are traditionally difficult to solve using conventional computational methods. Specifically, it explores the application of PINNs to tackle the sensor placement problem. This is project is an implementation of Physics-Informed Neural Networks for Heat Transfer Problems

The primary goal of this project is to investigate and implement the use of Physics-Informed Neural Networks (PINNs) to solve complex inverse problems related to heat transfer. By harnessing the power of PINNs, the project aims to provide a fresh perspective that enhances the accuracy and efficiency of solving intricate heat transfer problems, which have traditionally posed challenges for conventional computational techniques.

• Implementation of Physics-Informed Neural Networks (PINNs) using the DeepXDE Python package.
• Solution of prototype heat transfer problems that prove challenging for traditional computational approaches.
• Application of PINNs to address the sensor placement problem.
• Identification of optimal sensor locations for capturing crucial information about a given system or environment.
• Detailed project report highlighting implementation details, methodology, and results.
• The code implementation is still a work in progress.

This project draws inspiration from the potential of Physics-Informed Neural Networks (PINNs) to transform heat transfer problem-solving. I extend my gratitude to the DeepXDE Python package for facilitating efficient and accurate PINN implementations.

For inquiries or collaborations, please contact Vishal Bondili.

Cai, S., Wang, Z., Wang, S., Perdikaris, P., and Karniadakis, G. E. (April 21, 2021). "Physics-Informed Neural Networks for Heat Transfer Problems." ASME. J. Heat Transfer. June 2021; 143(6): 060801.