Research On Multi-satellite Collaborative Planning And Autonomous Scheduling For Moving Targets Tracking

Moving target tracking has a wide range of application scenarios and significant practical value in military and civilian fields such as combating piracy crime,arresting drug smuggling,maritime disaster rescue,and space weapon early warning.Space-based remote sensing satellites,as the main space information collection platform,have the characteristics of wide coverage,long operation time,and freedom from national boundaries.These characteristics give space-based remote sensing satellites unique advantages in tracking moving targets.However,many uncertainties such as target motion variability,image recognition probability,and target arrival randomness also bring challenges to tracking moving targets by space-based remote sensing satellites.The new generation of agile satellites has flexible three-axis attitude maneuvering capabilities,which expands the observation window of the target;the traditional “ground planning + on-board execution” management and control mode is seriously lagging in response to moving targets,while the enhancement of on-board computing capabilities and the development of AI technology provide a prerequisite for on-board autonomous intelligence,which brings opportunities to solve the lagging shortcomings.Meanwhile,agile maneuverability and autonomous intelligence also raise new problems to the current satellite mission planning system: Agile maneuverability brings the constraints with time-dependent characteristics,making satellite planning and scheduling more complicated;moving targets have dynamic uncertainties,which require a more reasonable task management and task decision-making mechanism;timely response to moving targets depends on rapid,high-quality,and refined autonomous task scheduling;the tracking of multiple moving targets requires efficient inter-satellite collaboration.This research focuses on two types of moving targets: the marine low-speed moving targets(time-insensitive moving targets)and the space high-speed moving targets(time-sensitive moving targets).The research is carried out from the following aspects:First,the general solution framework for multi-satellite collaborative planning and autonomous scheduling problems for moving target tracking is studied.On the basis of introducing the target and resource characteristics,the problem of multi-satellite collaborative planning and autonomous scheduling is elaborated and analyzed,and a hierarchical modular general solution framework is proposed.The framework includes the on-board autonomous task management layer,the inter-satellite autonomous collaboration layer,and the on-board autonomous task scheduling layer.Each layer and the modules in the layer can realize the combination adjustment and sequence matching of the modules according to the collaborative mode of resources.Based on this framework,a centralizeddistributed collaborative structure is designed for time-insensitive moving targets;A decentralized collaborative structure with mutually exclusive target pools is designed for time-sensitive moving targets,and the mutually exclusive target pool is introduced to reduce communication costs.Secondly,on-board autonomous task management for moving targets is researched.On-board autonomous task management is divided into target motion prediction,autonomous task generation,and priority coordination,which is responsible for transforming the uncertainty of the target into tasks with quantitative attribute description.For timeinsensitive moving targets,a double-constrained prediction model based on Gaussian distribution is proposed,a decision tree for autonomous task generation is constructed,and a multi-level task priority coordination strategy is introduced.For time-sensitive moving targets,a target prediction model based on the combination of elliptical orbit and RungeKutta integral is adopted,and dynamic priority is introduced to realize the dynamic attribute configuration of the target.Thirdly,under the centralized-distributed collaborative structure,the single-satellite autonomous task scheduling problem and the multi-satellite collaborative task assignment problem for time-insensitive moving targets are studied.Aiming at the former problem,a time-attitude adjacency graph model is introduced to describe it.From the idea of dynamic programming and sequential solution construction,the constraint model of the problem is transformed into a Markov decision process(MDP)model.In problem solving,the graph attention network is used to extract the features of the problem,and the network is trained by deep reinforcement learning based on proximal policy optimization.The simulation results show that the algorithm not only surpasses heuristics,classic meta-heuristics,and deep Q-learning algorithms by a large margin in terms of solution quality,but also has a fast solving speed.Aiming at the latter problem,an integer programming model is established on the basis of the problem description.In the process of serialization decisionmaking,evolutionary rules are used to solve this problem by the idea of sequence solution construction,and a constructed heuristic method based on gene expression programming evolution is proposed.The simulation results reflect that the evolutionary rule completely surpasses three types of heuristic rules with adaptive mechanisms,and achieves a solution effect close to meta-heuristic algorithms.Additionally,the solving speed is also quite fast.Finally,under the decentralized collaborative structure with mutually exclusive target pools,the problem of multi-satellite autonomous collaborative task planning for timesensitive moving targets is researched.On the basis of analyzing the characteristics of time-sensitive target collaborative tasks,a specific design of a decentralized collaborative structure with mutually exclusive target pools is carried out.MDP is used to realize the modeling of single-satellite autonomous task scheduling,and the modeling of multi-satellite collaborative task planning is realized based on the decentralized Markov decision process(DEC-MDP).After analyzing the difficulty of solving the problem,a method based on knowledge rules is proposed to solve the single-satellite autonomous task scheduling problem.Then,for the inter-satellite online collaboration,a request-response based task planning conflict resolution mechanism is designed,and the maintenance and transfer strategies of the targets in the mutually exclusive target pool are considered to improve the overall profits of the system.The simulation results exhibit that the algorithm and mechanism can effectively solve the multi-satellite autonomous collaborative task planning problem for time-sensitive moving targets,and it has the practicality of on-board deployment.