Interplanetary trajectory optimization with differential evolution.

摘要

Global optimization methods have been increasingly under consideration for the preliminary design of high-thrust interplanetary missions. One promising global method, Differential Evolution (DE), has previously been included in two comparative studies that resulted in contradictory performances for the algorithm. Differences in the three DE tuning parameters, population size, crossover probability, and mutation coefficient, are hypothesized to be the cause of the conflicting results. By experimentally investigating the relationship between tuning parameter values and the quality of solutions obtained, DE is found to be very sensitive to the tuning parameter values. A small number of tuning parameter sets are identified that result in rapid optimization of a diverse group of typical ballistic interplanetary trajectories. The tuned DE routine has been implemented in a new tool, MDTOP (Mission---Direct Trajectory Optimization Program), and the results have been characterized over an array of challenging problems.