Research On Autonomous Task Planning Of Imaging Satellite Of Formation Flying

With the rapid development of imaging satellite technology in China, customers have proposed higher requests to the accuracy and rapidity of satellite earth observation. It has become a tendency that several imaging satellites cooperate with each other and work together, to execute earth observation tasks based on the formation flying of satellites. Traditionally, the sequences of the observation task for the multi-satellites formation are calculated in advance by the centralized ground mission center, then uploaded to every satellite to be executed. To realize the fast response of satellites to the real-time imaging tasks, the autonomous task planning technology has become a research hotspot. With this technology, imaging targets are distributed to satellites quickly by the on-orbit satellite formation, instead of the ground mission center.In this paper, oriented to multiple imaging targets in hot area, the autonomous task planning of a satellite formation has been analyzed. Firstly, an orbit recursion algorithm with J2 perturbation and a preliminary-judging method based on closed geometric graph are put forward, so as to process the data of satellites and targets preparatorily.Then considering every constraint involved in the observation process, some ground targets are selected to form a data set for every satellite, by the condition of being able to be imaged by this satellite when there is no existing target in its task sequence. Autonomous task planning algorithms are designed on the basis of these data sets.Firstly aimed at single imaging satellite, a hierarchic-unoptimized task planning algorithm is proposed based on the priorities of targets. Targets in hot area are divided into three groups by their priorities, which are important, secondary and ordinary. A taskdistribution with the process of “inputing targets- judging with constraints- arranging task sequence” as the core is put forward, then it is used for sifting the important, secondary and ordinary targets successively to form an observation task sequence for the satellite. Based on the fact that observation task sequence varies with the order of the targets input, a single-satellite hierarchical-optimized task planning algorithm is proposed. Within each hierarch, targets of the same group are input with multiple orders, the formed observation sequences with the current maximum imaging gain are selected. And the targets of the next hierarch will be distributed, then added to the vacancies in of the observation sequences of the previous hierarchy. Finally one or multiple observation sequences with the maximum imaging gain are formed. Then the single-satellite hierarchical task planning algorithm is expanded to the case of multi-satellites formation.The multi-satellites hierarchical task planning algorithm is based on the priorities and imaging preference of targets, with the targets distribution as a core. Within each hierarch, targets with the same imaging preference are successively input to every satellite in the satellite group to be distributed. In the group every satellite carries a device of the specific imaging preference. Targets not added to observation sequences will be input to the next satellite group, until they are inserted to observation sequence of a satellite, or all the satellites have been judged. Similarly, the multi-satellites hierarchical-optimized task planning algorithm is designed by inputting targets with multiple orders, then selecting the observation sequence with the maximum imaging gain by group and then by hierarchy. Finally the targets are fast distributed to satellites with a high imaging gain.Simulations are given to verify the single and multi-satellites hierarchical task planning algorithms respectively. The simulation of multi-satellites task planning algorithm is given with the background of 160 ground targets in Wenchuan, Si Chuan Province and ten satellites of formation flying. And Imaging gain and the number of targets in observation sequences are chosen as the optimization goal respectively, the results of the simulations verify the rapidity, validity and wide applicability of the algorithms proposed.