Abstract: he maneuver operations of certain imaging satellites are tightly constrained, which limits their observing activities. Task merging observation can improve satellite observing efficiency. A model of satellite observation scheduling with task merging is proposed in this article and a very fast simulated annealing algorithm (VFSA) is developed to solve the problem. Multiple neighborhoods are defined respectively for the atomic task and composite tasks, which achieve dynamic task merging and decomposing in a search procedure. With an adventure acceptable probability and fast annealing, VFSA can improve the convergence speed. Reannealing mechanism and three diversification strategies are defined to avoid local optimum solutions and exploit the larger space. Computation results demonstrate the effectiveness of the algorithm.

Key words: imaging satellite, task merging, scheduling, very fast simulated annealing algorithm, diversification strategy