A differential evolution-based optimization tool for interplanetary transfer trajectory design

The difficulty of optimizing interplanetary trajectory design problem is caused by extremely sensitive and highly nonlinear search space [3]. Figure 1 shows a highly nonlinear search space of Messenger (full) problem around the current known best solution. The real interplanetary exploration mission usually has to consider two basic objectives: minimum mission period and minimum fuel consumption. However, even though only one objective function is considered for mission optimization, the optimum is still very hard to be found. In 2005, the GTOP database published by ESA includes some interplanetary transfer trajectory design problems, which can be simply divided by multiple gravity-assist (MGA) problems and multiple gravity assists with deep space manoeuvre (MGA-1DSM) problems. In the GTOP database, MGA problem includes Cassini1 and GTOC1; MGA-1DSM problem includes Sagas, Rosetta, Cassini2, Messenger (reduced) and Messenger (full). With the given search space and special boundaries on each variable, the GTOP problems have shown big challenges on global optimization. Three aspects of the challenge can be described as follows: - The attraction basin of global optimum is rather smaller than those of other local optima, and unluckily, this optimal basin is usually hidden in the "not promising" local spaces. During the evolutionary process, it is so difficult to locate these not promising local spaces when considering the population motivation based on greedy gradient information. This causes a result that the general swarm intelligent algorithms can only find sub-optimal results, and the difference between global optimum and found local optima is not acceptable.