Boundary-Decoder network for inverse prediction of capacitor electrostatic analysis
This work addresses a domain-specific problem in computational electromagnetics by enabling faster simulations with changing boundary conditions, though it appears incremental as it builds on existing deep learning methods.
The authors tackled the problem of time-consuming electrostatic simulations that require re-solving or re-training when boundary conditions change, by proposing an end-to-end deep learning approach for inverse prediction. Their method significantly outperformed plain vanilla neural networks and physics-informed neural networks under dynamic boundary conditions while retaining forward modeling capability.
Traditional electrostatic simulation are meshed-based methods which convert partial differential equations into an algebraic system of equations and their solutions are approximated through numerical methods. These methods are time consuming and any changes in their initial or boundary conditions will require solving the numerical problem again. Newer computational methods such as the physics informed neural net (PINN) similarly require re-training when boundary conditions changes. In this work, we propose an end-to-end deep learning approach to model parameter changes to the boundary conditions. The proposed method is demonstrated on the test problem of a long air-filled capacitor structure. The proposed approach is compared to plain vanilla deep learning (NN) and PINN. It is shown that our method can significantly outperform both NN and PINN under dynamic boundary condition as well as retaining its full capability as a forward model.