Dynamically Distributed Genetic Algorithm and Its Application to Diversely Selectable Trajectory Optimization of Winged Rockets

Abstract

The development of an efficient and flexible guidance system is one of the most important aspects of studies on reusable space transportation systems such as winged rockets. A guidance system that can generate several good trajectories instantly is required for responding to sudden changes in trajectories that may lead to accidents. We therefore propose a flight trajectory optimization method that uses a dynamically distributed genetic algorithm (DGA). DGA optimizes solutions by dynamically dividing and merging the individuals using the convergence situation of population. In most conventional studies on distributed genetic algorithms, the population has been divided into a fixed number of groups. This constraint sometimes deteriorates the growth of the solution because the number of groups and individuals in each group is closely related to the convergence of the solution. Our proposed dynamically DGA, which uses hierarchical clustering, was verified by a computer simulation.