| As modern wars show increasing dependence on space information, it has become necessary to use satellite grouping in observation and detection when developing reconnaissance satellites. Aiming at efficient usage of satellite resources and satisfaction for users'demands, multi-satellite observation task-scheduling(MSOTS) has become an urgent problem to be solved.This paper tackles MSOTS in two aspects: model design and multi-objective optimization. In the first part, a model with both generality and compatibility has been developed, following the idea of seeking a standard model. In the second part, based on the notion of Pareto optimization, evolutionary algorithm is applied in solving the multi-objective optimisation problem(MOP), resulting in a scheme for the task-scheduling problem.Achievements in this paper is summaried are two aspects.1 On the basis of a thorough analysis of the imaging task, all the constraints are layered before they can be handled in different ways. Taking imaging tasks as objects, indentifying reconnaissance time windows to bags with multi-form characteritics, using a"compatible list"to describe the complex correspondence between objects and bags, a multi-demensional dynamic Knapsack model(MDDK) is achieved, which covers existing constraints and holds extensive properties for possible technology development in future. In this way MSOTS is summaried as a MOP.2 In solving the resulting MOP, a culture algorithm(CA) with multi-layered belief space is applied to MDDK, which is based on a pivotal statement of CA after an introduction of Pareto optimisation and a specification of difficulties in MOP. Thereafter, applying the seperation skill in MDDK, a solution scheme named BAG-CUL is developed for MSOTS.Simulation results show the effectiveness and computation efficiency of the scheme.The results of this paper provides a new direction for solving MOTS using standard models, as well as an instructive try in applying CA to MOP. |
PDF Full Text Request