Date of Award

2020

Document Type

Dissertation

Degree Name

Ph.D.

Department

Mechanical Engineering

First Advisor

Jason Roney

Second Advisor

Roman Herschitz

Third Advisor

Matthew Gordon

Fourth Advisor

Yi Yun-Bo

Keywords

Constellation, Genetic algorithm, Machine learning, Reinforcement learning, Satellite

Abstract

This paper will review results and discuss a new method to address the deployment and management of a satellite constellation. The first two chapters will explorer the use of small satellites, and some of the advances in technology that have enabled small spacecraft to maintain modern performance requirements in incredibly small packages.

The third chapter will address the multiple-objective optimization problem for a global persistent coverage constellation of communications spacecraft in Low Earth Orbit. A genetic algorithm was implemented in MATLAB to explore the design space – 288 trillion possibilities – utilizing the Satellite Tool Kit (STK) software developers kit. STK and MATLAB autonomously develop and analyze a variety of constellations by permutating altitude, inclination, number of satellites, number of planes, and Right Ascension of the Ascending Node (RAAN). The coverage of these constellations was calculated and evaluated utilizing a parametrically driven cost and revenue generation model to determine the most profitable constellation configuration.

The fourth chapter will discuss a novel method to address the optimization problem of ground station placement; enabling continuous communication with the mega constellation defined in the third chapter. A genetic algorithm implemented in MATLAB explored the globe utilizing Satellite Tool Kit to determining the optimal number of ground stations and their placement – considering local infrastructure available and the constellation connectivity during a 24-hour period. A new revenue-based fitness function evaluated these parameters and the potential revenue to determine the most lucrative configuration.

The final chapter will utilize the deep reinforcement learning algorithm Proximal Policy Optimization (PPO2) on a custom spacecraft build and loss model, to determine if an AI can learn to monitor the health of a constellation of satellites and develop an optimal replacement strategy. A custom environment was created to simulate how spacecraft are built, launched, generate revenue, and finally decay. The reinforcement learning agent successfully learned an optimal policy for two models, a simplified model where the cost of the actions was not considered, and an advanced model where cost was included as a major element. In both models the AI monitored the constellations and took multiple strategic and tactical actions to replace satellites to maintain constellation performance – ensuring there was never a shortage of satellites.

Publication Statement

Copyright is held by the author. User is responsible for all copyright compliance.

Provenance

Received from ProQuest

Rights holder

Joseph Ryan Kopacz

File size

143 p.

File format

application/pdf

Language

en

Discipline

Aerospace engineering, Computer science

Share

COinS