A Multi-Objective, Multi-Agent Transcription for the Global Optimization of Interplanetary Trajectories
- PDF / 2,219,400 Bytes
- 29 Pages / 439.642 x 666.49 pts Page_size
- 93 Downloads / 145 Views
A Multi-Objective, Multi-Agent Transcription for the Global Optimization of Interplanetary Trajectories Sean W. Napier1
· Jay W. McMahon2 · Jacob A. Englander3
© American Astronautical Society 2020
Abstract Distributed Spacecraft Missions present challenges for current trajectory optimization capabilities. When tasked with the global optimization of interplanetary MultiVehicle Mission (MVM) trajectories specifically, state-of-the-art techniques are hindered by their need to treat the MVM as multiple decoupled trajectory optimization subproblems. This shortfall blunts their ability to utilize inter-spacecraft coordination constraints and may lead to suboptimal solutions to the coupled MVM problem. Only a handful of platforms capable of fully-automated multi-objective interplanetary global trajectory optimization exist for single-vehicle missions (SVMs), but none can perform this task for interplanetary MVMs. We present a fully-automated technique that frames interplanetary MVMs as Multi-Objective, Multi-Agent, Hybrid Optimal Control Problems (MOMA HOCP). This framework is introduced with three novel coordination constraints to explore different coupled decision spaces. The technique is applied to explore the preliminary design of a dual-manifest mission to the Ice Giants: Uranus, and Neptune, which has been shown to be infeasible using only a single spacecraft anytime between 2020 and 2070. Keywords Distributed spacecraft missions · Global trajectory optimization · Multi-Vehicle Missions (MVM) · Multi-Agent Multi-Objective Hybrid Optimal Control Problems (MOMA HOCP) · Interplanetary · Outer-loop · Inner-loop · Non-dominated sort · Transcription · Genetic algorithm · Pareto front · Coordination constraints · Ice giants
Ph.D. Candidate, Colorado Center for Astrodynamics Research, University of Colorado, Boulder, CO 80309, USA. Sean W. Napier
[email protected]
Extended author information available on the last page of the article.
The Journal of the Astronautical Sciences
Introduction Framing interplanetary spacecraft trajectory optimization as a hybrid optimal control problem (HOCP) has proven an effective approach [2, 3]. In a HOCP framework, trajectory optimization is a Mixed-Integer Programming (MIP) problem. Some decision variables are discrete (integers) while others are continuous (floating point), necessitating distinct optimization routines for each type. Furthermore, the resulting mission designs are points within a solution space spanned by multiple objectives (i.e., minimum fuel versus minimum time of flight). Thus, in order to effectively characterize the solution space for a given mission design problem, a multi-objective HOCP framework is essential. However, while tools exist to solve interplanetary multi-objective HOCPs for a single spacecraft, no tool exists to do so for multispacecraft problems. Addressing the shortcomings of current approaches to MVM optimization, including the methods for handling inter-spacecraft coordination objectives and constraints, are key to enabling the opti
Data Loading...
