Research On Clustering Of Satellite Imaging Reconnaissance Tasks

Clusteringofsatelliteimagingreconnaissance tasks isaprocess that mergingsometasks which can be executed bythe same satellite intoa single task according to certainrules. Clustering of tasks can reduce solution space of tasks programming and improvesatellite usage efficiency,therefore ,it is very significance to research satellite imagingreconnaissance tasks clustering problem. IAnd there are few research reports about it.Based on analysis the characteristics of the satellite imaging reconnaissance tasksclustering problem, this thesis explores single-satllite iamaging reconnaissance tasksclustering problem and multi-satellite imaging reconnaissance tasks clustering problem,and establishes the clustering models and algorithms. It is validated in the applicationinstance.Themaincontentsofthesis isasfollows:(1)A method of clustering of single-satellite imaging reconnaissance tasks ispresented.Research on clustering restriction condition of spot target atomic tasks andpolygon target atomic tasks. According to atomic tasks whether satisfying clusteringrestriction condition, we establish graph theory model. Translate clustering ofsingle-satellite imagingreconnaissancetasks problem into clique partition problem ,anddesigncliquepartitionclusteringalgorithm.(2)Two clustering algorithms are proposed to solve clutering of multi-satelliteimaging reconnaissance tasks problem. Establish mathematic model for clustering ofmulti-satellite imaging reconnaissance tasks problem,and put forward heuristicclustering algorithm based on multi-resources matching atomic inserted and clusteringalgorithm based on compatible relation.The idea of heuristic clustering algorithm basedon multi-resources matching atomic tasks inserted is:dividing atomic tasks intosingle-resource matching atomic tasks and multi-resources matching atomictasks,single-resource matching atomic tasks clustering in the same way assingle-satellite imaging reconnaissance tasks clustering, multi-resources matchingatomictasksinsertedintothesingleorbittasks clusteringgraphinturn,and choosedthematching satellite and clustering scheme according to heruristic rules. The idea ofclusteringalgorithmbasedoncompatiblerelationis:firstly,searchmaximalcompatiblesubsets from every atomic tasks set tasks of which math the same satellite; secondlyfind clustering likely scheme of the multi-resources matching tasks according to themaximal compatible subsets; at last, choose the clustering scheme which consumesatelliteenergiesleast.(3)Application examples are studied. According to the solution of the applicationexample of single-satellite imaging reconnaissance tasks clustering problem and theapplicationexampleofmulti-satelliteimagingreconnaissancetasksclustering problem,itvalidates thattheclusteringmodelsandalgorithmsarefeasibleandeffective.