Physics Informed Neural Networks for Gas Turbine Applications

The physical world around us is profoundly complex, and for years, researchers have sought to develop a deeper understanding of how it functions. Building models capable of predicting the characteristics of multi-physics systems continues to be a critical challenge within the sciences. Gas Turbines (GTs) are a class of such systems where Deep Learning models have been applied to address some of the domain-specific challenges.

As powerful as they are, neural networks are purely data-driven. Physics Informed Neural Networks (PINNs), on the other hand, are data-driven and can abide by the equations that govern the underlying physics. Recently, there has been growing research in applying PINNs for predicting flow and heat transfer in relatively simple scenarios.

In these applications just by providing the boundary conditions and informing the neural network of the governing Navier-stokes equations, flow parameters are predicted without the need for any mesh or other numerical models. Such simple applications can be extended to industrial scenarios with significant scope for the application of PINNs in gas turbines.

This project aims to investigate the development of PINNs for GT applications and is an excellent opportunity to develop and apply state-of-the-art machine learning techniques to solve real-world problems.