Examenspresentation av Felix Ågren och Carl-Johan Björnson

Supervisors: Olof Hjortstam (Hitachi Energy Research) and Christian Häger (Chalmers)
Examiner: Yuriy Serdyuk
Opponents: Stina Torell and Elinor Einarsson

Abstract:
The development of efficient high-voltage equipment is imperative for minimizing greenhouse gas emissions and achieving cost savings within the energy system. Effective electrical insulation plays a pivotal role in such development and requires an understanding of the performance of gaseous insulators, such as air, under high-voltage stress. Electric discharges and charge transport in gases are typically modeled using systems of partial differential equations (PDEs) and their solutions are traditionally approximated numerically with discretizing methods such as the finite element method (FEM). However, such methods have significant shortcomings including difficulty of handling high-dimensional problems, non-smooth behaviors, and inverse problems with hidden physics.

An emerging, mesh-free alternative to traditional numerical methods is Physics-Informed Neural Networks (PINNs). Within this approach, a neural network is built and PDEs with associated constraints are embedded into the network's loss function. Initial experiments with PINNs for the forward problem of electric field driven charge transport in gas discharges have shown promising advantages compared to FEM, but failed to model strongly non-uniform concentration profiles of charge carriers and coupled electric field distribution within the domain. This thesis contributes to this research by showing how a variety of performance-enhancing techniques can address the weaknesses of previous works, improving accuracy and enabling the modeling of steeper gradients of transported space charges and associated electric field. Additionally, it is demonstrated how PINNs can be used to solve inverse problems related to electrical gas discharges and space charge dynamics, discovering both unknown model input parameters and distributions, in particular, mobilities of charged species.

Everyone is most welcome!

Felix, Carl-Johan and Yuriy