- Version
- Download 168
- File Size 483.62 KB
- File Count 1
- Create Date 23/11/2022
- Last Updated 23/11/2022
A Review Paper on Physics Informed Neural Networks (PINNs)
Rohit Chavan1, Pranay Nimse2, Shubham Patil3, Sharvari Dabhade4
Prof. Abhay Gaidhani
Department of Computer Engineering SITRC, Nashik
Abstract - Physics-Informed Neural Networks (PINNs) are neural networks (NNs) that encode model equations such as Partial Differential Equations (PDE) and physical laws as a component of the neural network. PDEs, fractional equations, integral-differential equations, and stochastic PDEs are now solved using PINNs. This novel methodology evolved as a multi-task learning framework in which a NN must fit observed data while decreasing a PDE residual. According to the study, the majority of research has focused on customizing the PINN using various activation functions, gradient optimization techniques, neural network structures, and loss function structures. Despite the wide range of applications for which PINNs have been used, advancements are still possible by demonstrating their ability to be more feasible in some contexts, most notably theoretical issues that remain unresolved.
Keywords: deep learning, neural network, power system dynamics, power flow, system inertia