Genetic Algorithms for Minimal Fuel Consumption of Electric Propulsion Space Vehicles Bandi Bharat Kumar Reddy, Abdollah Homaifar, and Albert C. Esterline Autonomous Control and Information Technology Center Department of Electrical and Computer Engineering North Carolina A&T State University, NC Email: {br018138, homaifar, esterlin } @ ncat.edu Abstract- This paper demonstrates the utility of genetic algorithms (GAs) to determine a near optimal control strategy for electric propulsion systems. The various strategies implemented are simple GA, simple GA with elitism and micro GA. The accuracy and performance of the control strategy obtained using these methods are discussed along with their detailed description. This work inherently validates the use of ionic thrust for deep space missions. 1. INTRODUCTION This paper demonstrates the utility of genetic algorithms to determine a near optimal control strategy for electric propulsion systems. The motivation behind this work is to traverse from Earth to the moons of Jupiter in the JIMO (Jupiter’s Icy Moon Orbiter) spacecraft. JIMO is scheduled to be launched in 2015, and it is projected that by then the technology will be available for the prolonged usage of ionic propulsion. In order to appreciate the use of ionic propulsion systems better, a brief design and plan of conventional orbit transfers is first overviewed. The Patched-Conic Approximation [Bate, R.R. et al, 1975 and Prussing, J.E and Conway, B.A., 1993] is a technique to analyze a mission consisting of a spacecraft and several celestial bodies as a two body problem, with one body always being the spacecraft. Here, the velocity and radius of the spacecraft at destination (target planet or any celestial body) is determined by three initial conditions of the spacecraft with respect to the Earth (namely, geocentric radius, initial velocity, and flight-path-angle of the spacecraft) and a final state, which is the angle that specifies the point at which the geocentric trajectory enters the destination’s sphere of influence. (A planet’s sphere of influence is defined as the region of gravitational influence around it). Even though patched-conic approximation is the approach most often used in analyzing orbital transfers, its high dependence on the initial velocity produces solutions that involve transportation of large amounts of fuel required to create the initial thrust while exiting the Earth’s sphere of influence. Moreover, the resulting high velocity is not always sufficient to reach the desired final orbit. A change from the lower orbit to the higher orbit and vice versa can also be achieved through the Hohmann transfer [Bate, R.R. et al, 1975, p-163]. Here the spacecraft is fired with a high thrust from the lower orbit to reach the desired higher orbit. On reaching the higher orbit, the spacecraft’s velocity is not only lower than in the initial lower orbit, but lower than what is required to maintain it in the final higher orbit. At this stage the spacecraft is fired again with a high thrust to continue in the higher orbit. Even though this change in speed is achieved through the double-tangent transfer ellipse, the energy required to transport the large amounts of fuel for these intermediate high thrusts is quite expensive. Another novel technique is to use gravitational assists from celestial bodies between the Earth and the target planet. With gravitational assists, the spacecraft’s motion is preplanned so that, in the proximity of the intermediate planets, the craft’s speed is either increased or decreased, and its heading direction can be changed. As this method depends heavily on the location of the intermediate planets at the spacecraft’s time of launch and requires substantial mission planning, there is little or no scope for mission alterations. In comparison to these methods, an ionic propulsion system has a flexible control strategy, and, when coupled with the use of intermediate thrusts (thrust ON and OFF), it becomes a viable alternative to the conventional methods discussed so far. This intermediate thrust sequence can be obtained by using the bang bang control strategy for minimal fuel problems [Kirk, D.E.,1970, pp 260-284]. Determining this control strategy using bang bang control, however, is very complex for our problem, which addresses a highly nonlinear system. By using a GA, the overall complexity of the problem is drastically reduced and solved with considerable simplicity. Another advantage of using a GA over other heuristic methods is its relative simplicity in modeling the overall control strategy for this mission (Chromosome Representation). 2. PROBLEM DEFINITION The objective of this research is to compare the control strategies obtained by simple a GA, simple a GA with elitism and a micro GA with elitism for the JIMO spacecraft to consume minimal fuel while traversing between two points in space represented by longitude, latitude and radius. The control profiles of the spacecraft are thrust ON (thrusting), thrust OFF (coasting), thrust angle of attack and Proceedings of the 2005 International Conference on Computational Intelligence for Modelling, Control and Automation, and International Conference on Intelligent Agents, Web Technologies and Internet Commerce (CIMCA-IAWTIC’05) 0-7695-2504-0/05 $20.00 © 2005 IEEE