Multi-Objective Evolutionary Algorithms for Low-Thrust Orbit Transfer Optimization

We address the problem of optimizing a spacecraft trajectory by using three different multi-objective evolutionary algorithms: i) Non-dominated sorting genetic algorithm, ii) Pareto-based ranking genetic algorithm, and iii) Strength Pareto genetic algorithm. The trajectory of interest is an orbit transfer around a central body when the spacecraft uses a lowthrust propulsion system. We use a Lyapunov feedback control law called the Q-law to create an eligible trajectory, while the Q-law control parameters are selected with the multiobjective algorithms. The optimization goal is to minimize flight time and consumed propellant mass simultaneously. The Pareto fronts (trade-off surface between flight time and propellant mass) produced by these algorithms are evaluated by means of two quantitative metrics: 1) size of the dominated space and 2) coverage of two Pareto fronts. With the two metrics, a hierarchy of algorithms emerged. The nondominated sorting genetic algorithm and the strength Pareto genetic algorithm are equally effective, and they outperform the Pareto-based ranking genetic algorithm.