考虑多类型时间依赖资源约束的敏捷卫星调度优化

随着现代空间科技的迅猛发展, 光学遥感图像数据的应用需求越来越广泛, 大力推动了光学对地观测卫星的发展. 然而, 由于高昂的发射成本的约束, 对地观测卫星的资源是有限的, 远远无法满足各类数据需求. 因此, 提高对地观测卫星的使用效率, 提高其任务执行率, 具有非常重要的应用价值. 本文聚焦于敏捷对地观测卫星的任务调度问题, 即在给定的调度周期内, 对有限的卫星资源制定合理的任务调度方案, 在满足一定星上资源约束下, 最大化观测任务收益. 该问题难点在于星上的资源是非常有限的, 例如存储图像数据的固存资源、用于采集数据和卫星姿态切换的能量资源及执行任务活动耗费的时间资源. 需要注意的是, 能量消耗量和时间消耗量依赖于任务的执行时间, 这是敏捷卫星相对传统的非敏捷卫星独有的特性. 不同任务场景对不同类型资源的需求不同, 多种资源约束互相耦合, 资源约束具有时间依赖特性, 这些难点无疑极大地增加了卫星调度的求解难度. 为高效地求解该问题,本文构建了多类型时间依赖资源约束的敏捷卫星调度整数规划模型, 并针对问题特性提出了一种基于自适应选择因子的迭代局部搜索启发式算法. 自适应选择因子综合考虑了目标收益、资源消耗量、资源约束的松弛量, 采用动态变化的资源重要度, 能快速自适应地根据当前场景下各种类型的资源数据使用量来确定最佳局部搜索方向, 从而在有限时间内找到高质量的解. 实验结果证明, 本文所提出的算法在多种情况下相比当前最好算法求解效果显著更优. 此外, 算法独有的自适应选择因子相比传统的选择因子的求解质量更高, 这是因为所设计的自适应选择因子兼顾了目标收益和资源消耗量之间权衡关系的同时, 采用动态变化的资源重要度准确捕捉了资源需求的迫切程度.

The agile earth observation satellite scheduling with multiple resource constraints

With the development of space technology, the demand for optical remote sensing image is becoming increasingly widespread. However, due to the high cost of launching, the resources of satellites cannot meet the needs. Therefore, improving the efficiency of satellite utilization is of great value. This article focuses on the scheduling problem of agile ob-servation satellites, which involves developing scheduling plans within a given scheduling period under on-board resource constraints. The difficulty lies in the limited resources, such as fixed storage resources for image data, energy resources for data acquisition and satellite attitude switching, and time resources. Energy consumption and time consumption depend on the observation time, which is exclusive for agile satellites compared to non-agile satellites. Different scenarios have different demands for different types of resources, multiple resource constraints are coupled, and resource constraints are time-dependent. These challenges greatly increase the difficulty. This article constructs an integer programming model and proposes an iterative local search heuristic based on an adaptive selection factor. The factor considers target benefits, resource consumption, and resource constraints. The algorithm can adaptively determine the local search direction based on the usage of various types of resource data in the current instances, thus finding high-quality solutions. Experimental results demonstrate that the proposed algorithm outperforms the current best algorithm. Furthermore, the adaptive selection factoryield high solution quality because the factor consider the trade-off between target benefits and resource consumption, and accurately capture the urgency of resource demand.