Research On Remote-Sensing Satellites Distributed Scheduling Model And Algorithm

With the development of our nation's aerospace projects, the number of satellites traveling on orbits is increasing. As an important type of Earth Observation Satellites, the function of Remote-Sensing Satellites (RSS) was improving, along with the increasement of the scale. As a result, Earth's-Observing request has been more complex and the real-time requirement of the scheduling of RSSs on ground was unrelenting advancing. Thus, How to establish a better system for scheduling of RSSs on ground, which can finely satisfy the user's request as well as has extendibility, was of great theoretical and practical value both in military and civilian areas. Based on the trend of current computing, this dissertation studied the scheduling of RSSs on ground problems in the field of distributed computing. The main contributions of this dissertation can be concluded as the following four parts:(1) A distributed framework model of scheduling of RSSs was established based on the MAS (Multi-Agent System). Considering the characteristics and constraints in the signal collecting process and data transmission process of RSSs, satellites and resources of date transmission was modeled as an Agent. The distributed framework model, which can be extended easily, was able to effectively accelerate development cycles when new satellites and resources of date transmission were expanded into the system. This kind of designment accelerate development cycles when a new satellite or a new resource of date transmission joined the system.(2) A distributed algorithm for observing tasks scheduling of RSSs based on improved Contract Net Protocol is designed and implemented. The distributing process of the observing tasks was considered as the bidding part in the Contract Net Protocol. In the same time, traditional Contract Net Protocol was optimized by analyzing the process of observing tasks scheduling of RSSs. Experimental results show that: Compared with centralized algorithm, both efficiency and real-time of the distributed algorithm proposed were elevated.(3) A distributed algorithm for resources of date transmission scheduling of RSSs is designed and implemented, based on the combination of the cooperative evolution and distributed genetic algorithm. The optimization was acquired by cooperative evolution of sub-populations. Experimental results show that: the distributed algorithm worked well under dynamic condition, meantime, efficiency of the distributed algorithm proposed were enhanced compared with centralized algorithm.(4) According to the research works presented above, a scheduling and planning experiment system, which has friendly multi-view interfaces of interaction between human and machines, based on C/S architecture for RSSs was designed and implemented as the platform to verify our research achievements. The system provided services, including scheme displaying, scheme editing, agents managing and so on.