Research On Multi-Task Rapid Scheduling Technology For Satellite Networks

With the current economic development and continuous advances in science and technology,competition in the space has become the frontier of research and development in various countries.Satellites,as an extremely important component in the exploration of airspace,can provide important support for the consolidation of national defense and the development of the national economy.At present,satellites have been developed for national defense security,agricultural construction,rescue and disaster relief,and weather forecasting,enormous contributions.In the future development of the society,the application of satellites will be more extensive,and it will become even more important to play a role.With the increasing size of satellite networks and the increasing number of satellite types,past single-star planning or task scheduling under small-scale networks has no longer adaptable to the current state of development.How to dispatch large amounts of satellite network resources and provide better services to more users has become the focus of current research in various countries.In this context,this paper designs a dynamic and adjustable Multi-Agent mission planning system method,which can effectively realize the multi-task rapid planning of satellite network.Based on this,a satellite mission planning software platform is designed.The research content of this article is as follows: 1.For the problem of task conflict and resource waste caused by the mismatch between task requirements and satellite resources,a satellite resource clustering algorithm based on task resource demand granularity matching is designed so that the satellites in the satellite group have the resources matching requirements of tasks',referring to the clustering method of machine learning.Referring to the distributed system framework,a distributed Multi-Agent system is designed to solve the multi-satellite mission planning problem: The satellite that matches the granularity of mission resources is constructed as an agent and tasks are assigned to the corresponding satellite agents.In each Agent,the mission planning problem was modeled by analyzing the mission requirements,the satellite constraints,and the on-board task conflict relationships.Based on the analysis of task-to-task conflict relationship on the satellite,a task execution conflict graph is established.Based on this,a greedy algorithm based on the shortest path in graph theory is designed to realize the task planning in the subagent.Therefore,distributed parallel planning can be performed through the Multi-Agent architecture,which greatly improves the efficiency of task planning.2.A dynamic and adjustable multi-agent task scheduling algorithm(DAMTSA)is designed for the performance loss caused by the planning method of satellite clustering and partitioning agent.The algorithm has an agent resource adjustment mechanism based on cooperative game,which enables different agents to cooperate with each other and re-allocate their respective resources,thereby making the resource allocation in the agent more reasonable,and the resource utilization rate and task completion rate higher.At the same time,the algorithm realizes the re-use of the satellite resources in each agent through designing a bidding method and contract mechanisms,further improving the task completion rate and resource utilization rate.At the same time,it also enables the planning system to cope with emergency missions that arrive dynamically.3.According to the actual requirements of current mission planning,this paper designs a satellite mission planning software platform.The software platform can realize graphical operations,facilitating the acceptance of user-submitted tasks.After the acceptance of the tasks,the software platform can schedule ths tasks using the task planning method.Referring to the less consideration of task scheduling correctness verification and lack of dynamic demonstration effects of current mission planning,the system software can perform 2D/3D dynamic demonstration of the planning results,which can better verify and demonstrate the results of task planning algorithm execution.Considering the extensibility of the mission planning platform,through the agreement of the mission planning algorithm interface,the software platform can support the invocation of different planning algorithm programs for mission planning and the comparison and display of results.