Physics informed neural network for charged particles surrounded by conductive boundaries

TL;DR

This paper introduces a physics-informed neural network (PINN) model to accurately predict potentials of charged particles near conductive boundaries, outperforming standard ML models and reducing computation time.

Contribution

The paper presents a novel PINN-based approach for modeling electrostatic potentials around charged particles with conductive boundaries, achieving high accuracy and efficiency.

Findings

Mean square error < 7%

R2 score > 90% for PINN

Significant reduction in computation time

Abstract

In this paper, we developed a new PINN-based model to predict the potential of point-charged particles surrounded by conductive walls. As a result of the proposed physics-informed neural network model, the mean square error and R2 score are less than 7% and more than 90% for the corresponding example simulation, respectively. Results have been compared with typical neural networks and random forest as a standard machine learning algorithm. The R2 score of the random forest model was 70%, and a standard neural network could not be trained well. Besides, computing time is significantly reduced compared to the finite element solver.