EP_DE II: A significant algorithm to search the optimal solution for global optimization of multi-gravity assist trajectory

Abstract

The paper deals with the design of the optimal multiple gravity assist trajectories. An improved search algorithm named EP_DE II with search space exploration, principal component analysis, guarantee mechanisms, and processing methods of search space is proposed based on EP_DE. First, a parameter is employed to start retaining the boundary information of the whole population. When the number of generations in evolutionary process reaches to this parameter, distribution range of population will be retained for the later computation. The best solutions in each generation are also recorded until the end of EP_DE II. Then the principal component analysis was conducted to find a cutting dimension for local search space, and selection process of cutting points are directed by the stored information before. Finally, search process is performed in all partitions of the search space. Global search experiments concerned about Benchmark Cassini1 and fly-by sequence EVVEEJS are presented to prove the efficiency of EP_DE II algorithm, comparing with basic differential evolution algorithm and EP_DE algorithm.