Research On Autonomous Agile Earth Observation Satellite Scheduling Algorithm

The scheduling problem of an Agile Earth Observation Satellite(AEOS for short)consists of selecting and scheduling a subset of targets among a set of candidate ones that satisfy imperative constraints and maximize a function of total profit.The constraints contain the visible time window constraint,the transition time constraint of attitude maneuver,the solid state memory constraint and the electricity constraint.The maneuvering capacity of the pitch axis makes the AEOS could image the target not under the star point.It increses the visible time windows between the satellite and targets,however,it also increses the complexity of solving the scheduling problem because of the larger solution space.Another point complicating the scheduling problem of AEOS is the time-dependent character of the transition time constraint of attitude maneuver.So the key constraint of the AEOS scheduling problem is the time-dependent temporal constraint composed of the visible time window constraint and the transition time constraint.In this dissertation,the time-dependent temporal constraint of AEOS scheduling problem is studied at first,and the research on the algorithms of autonomous AEOS scheduling problem in different scenarios are gradually developed.The main contributions of the dissertation are as follows:(1)A branch and bound algorithm is developed to solve the AEOS scheduling problem with the time-dependent temporal constraint.To solve the AEOS scheduling problem which integrates a time-dependent temporal constraint,we propose a highly efficient branch and bound algorithm whose effective ingredients include a look-ahead construction method for generating a high quality initial lower bound,and a combined use of three pruning strategies that help to prune a large portion of the search space.The experimental results showed that the proposed algorithm is efficient enough for engineering application.Furthermore,we carried out additional experiments to analyze the contribution of each key algorithm ingredient.And the research of the branch and bound algorithm also lays the foundation for the following research work of the thesis.(2)A bi-satellite cluster is introduced for the scenario of targets recognition over sea,and an online scheduling algorithm is proposed for the AEOS of the cluster.In recent years,it has becom the focus of many subjects how to effectively improve the efficiency of the sea search and rescue.To fulfill the mission of targets recognition over sea,we consider a bi-satellite cluster composed of an autonomous low resolution satellite(LRS)leading the formation for targets detection and a trailing agile high resolution satellite(HRS)for targets recognition.We focus on developing a method that is able to generate a schedule plan onboard the HRS taking into account the information received from the LRS,which amounts to solving an AEOS scheduling problem.Experimental results on a set of representative scenarios show that the proposed algorithm is effective which promotes significantly the bi-satellite cluster to improve the efficiency of targets recognition over sea.(3)An onboard decision model to slect the observation target is developed,and the method to train the model via the ground computing resources is developed.To further improve the observational capacity of the autonomous AEOS over long periods,an onboard decision model is developed for the AEOS to select the next observation target in the condition that all the constraints,including the temporal constraint and the resource constraints,are satisfied.In general,the onboard scheduling algorithms have to make a decision to select a target based on the limited information such as the approaching targets.The “myopic” decision will cause the satellite to deplete the memory and electricity resources prematurely,and thus make the satellite have to abandon the observation of the high profit target due to the lack of observation resources.The onboard decision model proposed in this thesis could help satellites to make a more visionary decision,and the model could be used to complete the learning training in the ground segment with the satellite operation history data.With the help of the decision model,the satellite could coordinate the global guidance information of the historical data in each decision-making,avoid consuming the observational resources prematurely and improve the total observation profit of the satellite in the scene.Finally,the validity of this method is verified by simulation experiment.(4)The constellation structure of autonomous AEOSs,and the onboard cooperative scheduling algorithms of muti-AEOS is studied.We analyze the constellation structure of autonomous AEOSs according to the current satellite platform technology,and a distributed-centralized satellite constellation structure is proposed.The constellation structure could improve the dynamic response capacity of the whole constellation in the condition that the satellites take account of their conventional observation targets.And then,we researched on the cooperative allocation strategy of the constellation to the dynamic emergency targets.Ten different cooperative allocation strategies are analyzed by the simulation experiment.At last,we extract several characteristic features of the conventional observation targets of each satellite and use a support vector machine,which could select an appropriate strategy according to the features,to further optimize the cooperative allocation strategy.