Automation of Spacecraft Layout Using Evolutionary Optimization Methods

摘要

Satellite layout optimization design (SLOD) problem is proposed as an automatic approach to generate better layout candidate schemes for engineers to choose from. It is addressed as a three-dimensional packing problem with behavioral constraints, which has been proven as a NP-hard problem theoretically. The goal of optimization is non-intersection of components, minimization of deviations of mass centre and axes of inertia. To solve this problem, it is proposed to use a two-epochs genetic algorithm (GA) with polishing. In the first epoch, bearing faces for component and rotation are selected. In the second epoch, location coords on the bearing face are selected. And in the polishing, where Nelder-Mead method is used, location coordinates are corrected.