elib
DLR-Header
DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Multiphase Low-Thrust Trajectory Optimization Using Evolutionary Neurocontrol

Ohndorf, Andreas (2016) Multiphase Low-Thrust Trajectory Optimization Using Evolutionary Neurocontrol. Dissertation, Technische Universität Delft.

[img] PDF
7MB

Abstract

To fulfill the objectives of deep space missions, such as in situ measurements at an outer planet’s moon or investigations at main belt asteroids, spacecraft must be provided with sufficient energy to get to these distant objects. This energy, expressed with the so-called ∆V -budget, which is the sum of required velocity changes along a spacecraft’s trajectory. As today’s and future deep space missions are infeasible using chemical propulsion alone, their trajectories involve one or more close flybys at mass-rich celestial bodies to gain additional orbit energy. These maneuvers are called gravity assists and depend on the relative positions of the assisting body and the respective target. Due to the orbital motion of both bodies, the required constellation may however repeat only every few decades. This constrains both trajectory and mission design and small launch windows can be the result. Any project delay hence threatens an on-time launch and thus potentially puts an entire mission at risk. Mitigation of that risk is possible through using low-thrust propulsion which can provide the required ∆V of a deep space mission without gravity assists. Contrary to chemical propulsion, having thrust values up to kilo-Newtons at specific impulse (Isp) values of 300-400 s, low-thrust propulsion currently offers only approximately one Newton at maximum. This thrust is achieved either through the ejection of carried-along particles, which are accelerated to very high velocities, or the reflection of sunlight photons. High exhaust gas velocities and very low propellant consumption make the respective Isp of low-thrust propulsion one magnitude higher than for chemical propulsion. The low-thrust propulsion concept of solar sailing even utilizes the solar radiation pressure for the generation of thrust, making it independent on any propellant. The different characteristics of low-thrust propulsion and chemical propulsion result in different trajectories. Therefore the methods for the optimization of trajectories of chemically propelled spacecraft are of limited use for the optimization of low-thrust trajectories. New methods were developed for this purpose, and one of them is Evolutionary Neurocontrol. This global optimization method combines the two biology-inspired mechanisms artificial neural networks and evolutionary algorithms. Called neurocontrollers, the artificial neural networks are used for spacecraft control. The optimization capability of evolutionary algorithms is used for the training of neurocontrollers. Contrary to other optimization methods, Evolutionary Neuroncontrol does not require an initial guess solution to work, which increases its usability for non-experts in optimal control and optimization. Evolutionary Neurocontrol was applied successfully in the past to various low-thrust transfer problems. Each of those problems, however, consisted of only one single heliocentric transfer from one celestial body to another. The problem of global optimization of multiphase low-thrust trajectories remained unsolved. This thesis describes how Evolutionary Neurocontrol can be extended to multiphase low-thrust transfers. An existing implementation was revised and complemented with new capabilities, concepts, and functionalities. Examples of the new features are a generic multiphase simulation framework, the support of nonheliocentric transfers, and third-body perturbation. The resulting method has been validated on various complex low-thrust transfer problems, which included two-phase transfers, like Earth-Moon-transfers, or heliocentric rendezvous missions with multiple targets or multiple propulsion technolgies. If available, the results were compared with published reference solutions. Finally, Evolutionary Neurocontrol was successfully applied to the design of a trajectory for a so-called Interstellar Heliopause Probe mission. Including a close flyby at Jupiter and using two different propulsion technologies, the resulting transfer brought the spacecraft to a heliocentric distance of 200 AU in less than 25 years.

Item URL in elib:https://elib.dlr.de/110672/
Document Type:Thesis (Dissertation)
Title:Multiphase Low-Thrust Trajectory Optimization Using Evolutionary Neurocontrol
Authors:
AuthorsInstitution or Email of AuthorsAuthors ORCID iD
Ohndorf, AndreasAndreas.Ohndorf (at) dlr.deUNSPECIFIED
Date:7 June 2016
Journal or Publication Title:Bibliothek der Technischen Universität Delft
Refereed publication:Yes
Open Access:Yes
Gold Open Access:No
In SCOPUS:No
In ISI Web of Science:No
Number of Pages:168
Status:Published
Keywords:Spacecraft Trajectory Optimization, Neural Networks, Genetic Algorithms, Low-Thrust
Institution:Technische Universität Delft
Department:Aerospace Engineering
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Space
HGF - Program Themes:Space Technology
DLR - Research area:Raumfahrt
DLR - Program:R SY - Technik für Raumfahrtsysteme
DLR - Research theme (Project):R - Raumflugbetrieb / Missionstechnologie
Location: Oberpfaffenhofen
Institutes and Institutions:Space Operations and Astronaut Training > Mission Operations
Deposited By: Schneider, Beatrice
Deposited On:13 Mar 2017 09:38
Last Modified:31 Jul 2019 20:07

Repository Staff Only: item control page

Browse
Search
Help & Contact
Information
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.