10.2514/1.A34043">
 

Multi-objective parallel optimization of geosynchronous space situational awareness architectures

Document Type

Article

Publication Date

11-2018

Abstract

This research explores a parallel implementation of a genetic algorithm (GA) to optimize geosynchronous orbit (GEO) Space Situational Awareness (SSA) architectures. It does so by parallel evaluation of executable (simulated) architectures on a high-performance computer (HPC). This effort had two primary objectives. The first was to develop and validate a robust methodology for design optimization of space constellations that avoids the typical limitations of point solutions, performance approximations, or limited design parameters. The second objective was to determine the near-optimal GEO SSA solution across multiple realistic objectives. The GA was implemented in Python, and architectures were modeled and simulated with AGI’s Systems Tool Kit (STK)™ on a 4600-node high-performance computing cluster. Results show that the GA finds increasingly “good” solutions in a 28-dimension design space and across 3 objectives, and these solutions were examined for sensitivity to preferences. After nearly 320,000 architectures were modeled, simulated, and evaluated on the HPC (27 years of CPU time, 3 days clock time), the near-optimal “maximum performance” solution consisted of 19 globally distributed telescopes, 4 satellites in equatorial low Earth orbit, and 4 satellites in near-GEO. The solutions obtained by this research outperform the extant literature and provide insight into large-scale, multidomain, multi-orbit disaggregated constellation designs and specifically GEO SSA solutions.

Comments

This article is published by AIAA and is available by subscription or purchase at the DOI link below.

Source Publication

Journal of Spacecraft and Rockets

This document is currently not available here.

Share

COinS