Electric Thrusters- Combining Particle-in-cell plasma simulations and Machine Learning

Abstract

The aim of the present project is to combine the Particle in Cell (PIC) approach with Machine Learning to advance the study of electric propulsion for spacecraft. This effort will focus on two concepts of thruster of special current interest- the famous Hall and the more frontier reconnection thrusters that received considerable attentions in recent years. This research will make first principle full PIC simulations in both two-dimensions and three-dimensions. This study will focus, for the Hall thruster, on the Landmark benchmark and for the reconnection thruster on our past investigations. Once satisfactory PIC simulations have been generated, machine learning will be utilized to discover the equations that can best model the behavior using simpler fluid models. To reveal these equations, a neural network approach will be used in tandem with sparsity-based model-discovery techniques. These can lead to the “correct� closure relations to be used to capture the essence of the full kinetic approach in a reduced model.

Document Details

Document Type
DoD Grant Award
Publication Date
Feb 22, 2024
Source ID
FA86552317030

Entities

People

  • Giovanni Lapenta

Organizations

  • Air Force Office of Scientific Research
  • Katholieke Universiteit Leuven
  • United States Air Force

Tags

Fields of Study

  • Physics

Readers

  • Aerospace Propulsion Engineering.
  • Computer Vision.
  • Theoretical Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks
  • Space
  • Space - Hall-Effect Thruster