Date of Award

Fall 2021

Publication Type

Thesis

Degree Name

M.A.Sc.

Department

Mechanical, Automotive, and Materials Engineering

Keywords

LSTM, Prognosis, Reaction wheel, Remaining useful life, Satellite

Supervisor

A. Rahimi

Supervisor

A. Cherniaev

Rights

info:eu-repo/semantics/openAccess

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Abstract

Artificial satellites are objects or a body that are stationed in the orbit of another object. The purpose of artificial satellites includes monitoring, information transfer, studying a different planet, space exploration, and fulfilling many other modern-day needs. For the increased demand, the number of artificial satellites revolving around the earth is also increasing. Due to cost efficiency, bulk manufacturing capability, and ease to launch in the orbits, small satellites are the topic of interest. Reaction wheels are widely used in the attitude control system of small satellites. Unfortunately, reaction wheels failure restricts the efficacy of a satellite, and it is one of the many reasons that lead to premature abandonment of the satellites. In larger satellites, there is room for mechanical redundancy to increase service reliability, so an onboard health monitoring system is in demand to ensure seamless performance by minimizing the risk factor of the sudden failure of a small satellite. This study observes the measurable system parameter of a faulty reaction wheel to estimate the remaining useful life of the reaction wheels. In this research, a data-driven approach is for the fault prognosis of the satellite reaction wheel. The measurable system parameters from the satellite reaction wheel are not directly related to the health of the system. So, the proposed method involves three stages to achieve the goal. In the first stage, the necessary observable system parameters are identified, and their future state is predicted based on historical data using a long short-term memory recurrent neural network. A health index parameter is defined and estimated using a multi-variate long short-term memory network in the second stage. In the third stage, the remaining useful life of the reaction wheel is estimated based on historical data of the health index parameter and a threshold. The approach is very efficient depending on the fault severity and can be used in on-field scenarios. The approach is robust up to a certain degree of noise, disturbance, and missing data.

Share

COinS