Performance Improvement of Hierarchical Fitness Calculation for Optimal Flight Trajectory Generation Using Dynamically Distributed Genetic Algorithm
Satoshi HIRAKIDA, Koichi Yonemoto, Takahiro Fujikawa, Masatomo Ichige, Kazuki KURIMOTO · The Proceedings of Conference of Kyushu Branch · 2019
In recent years, for further promotion of space development, reusable space transportation system with low launch cost per one has been researched and developed. However, there are some technical problems that must be solved in order to realize such system. One of these problems is the autonomous guidance system that can work under various flight environments. Therefore, onboard real-time optimal flight trajectory generation methodology using dynamically distributed genetic algorithm (DynDGA) has been studied. Its advantage is that it can cope with sudden changes in flight environment and mission by obtaining multiple optimized trajectories with diversity. While the general framework of this guidance system has almost been constructed, there is a problem that there are some performance indices of different importance and units, but the past algorithm sums up the raw values of performance indices with weight factor. Therefore, it was difficult to set appropriate weight factor such that fascinates the growth of the individual. In this paper, we propose a hierarchical fitness calculation method by fuzzy membership function so that performance indices grow according to priority order, and the performance of this guidance system is evaluated by numerical simulation.