The Study On Key Technologies Of Multiple Types Of Earth Observing Satellites United Scheduling

Earth Observing Satellites (EOSs) obtain information of the earth's surface from outer space by using satellite sensors. They are widely used in the social, economic, and military applications. In recent years, as more and more EOSs are launched to meet the needs of various complicated types of remote sensing data, the challenge to scheduling of multiple types of EOSs is increased. However, current researches focus on single type of satellite and simple requirements, which considers relatively simple earth observing procedure and restrictions, and lacks of methods and techniques for scheduling of multiple types of satellites, satisfying multiple types of requirements and combining satellite observing with data transmitting.To satisfy the united scheduling requirements of multiple types of low orbit satellites, including optical satellites, SAR satellites, electronic satellites, and ocean surveillance satellites, this dissertation studies the key technologies of the united scheduling of multiple types of EOSs. Research focus is put onto combining multiple types of EOSs with multiple types of data transmitting resources, thus to meet the needs of various earth observing tasks. The main work and contributions of this dissertation can be concluded as the following five parts:(1) We build a hierarchical description model for united scheduling of multiple types of EOSs. Due to the complexity of the united scheduling of multiple types of EOSs, the dissertation systematically analyzes and summaries the scheduling features in the target problem. Focusing on the characteristics of the scheduling workload and the diversity of the observing tasks, the dissertation concludes the problem model into five levels, and put focuses on four levels: resource constraints, scheduling constraints, business component and business process, meanwhile gives detailed description of the modeling process. The dissertation uses a component based modeling methodology to build the description model. Therefore, both the constraints and the optimization target are taken as components in the model. In the scheduling process, it is convenient to choose suitable process according to different characteristics of the observing tasks, the optimization strategy and the scheduling features. Moreover, during each process, the constraints and the optimization target functions can be combined dynamically according to different problem features. All of which make the description model highly compatible and extensible, thus could satisfy the complicated requirements of the united scheduling of multiple types of EOSs .(2) We propose the techniques to the scheduling of various data transmitting resources from both the satellites and ground stations. Results of the scheduling of data transmitting play an important role in satellite scheduling; therefore it must be studied first. Considering the variation of the on-satellite memorizer during the earth observing, the dissertation establishes a scheduling model for data transmitting resources based on the constraint graph, which transforms the scheduling problem into a multiple paths searching problem. By using the permutation based representation, an adaptive neighborhood searching mechanism is introduced to improve the neighborhood searching process. Moreover, a simulated annealing algorithm based on the adaptive neighborhood searching is proposed. Analysis to the convergence of the algorithm and the experimental results verify that the proposed model and algorithm are superior. Compared to other algorithms, the proposed method synthetically considers the following factors: the distribution of the earth observing tasks, the capability of the satellites' load, the effects of the real time transmission and storage data transmission to the scheduling. Therefore, the method could satisfy the data transmitting requirements more suitably. Moreover, compared to the simulated annealing algorithm, the stochastic hill-climbing algorithm and the genetic algorithm, the proposed method maintains much better global convergence.(3) We present a hybrid algorithm based on the GRASP (Greedy Randomized Adaptive Search Procedure, GRASP) framework for Benefit First strategy. The united scheduling problem has the similar characteristics with both the PDPMTW (Pickup and Delivery Problem with Multiple Time Windows) and oversubscribed scheduling problem. By using the mechanism of the repair search, the dissertation proposes a new hybrid algorithm which is based on the GRASP framework. The large neighborhood searching algorithm is introduced in the building phase. The simulated annealing algorithm is used at both the preservation of LNS (Large Neighborhood Search, LNS) results and each step of iterative search. The proposed method also presents a new iterative repair operator by combining the three operators of PDPTW, which has been proved to improve the search procedure effectively. Algorithm analyzing verifies the convergence of proposed method, moreover, a great deal of experimental results also show that the new hybrid algorithm maintains better global convergence.(4) We propose a new algorithm named HIGA (Hierarchical Immune Genetic Algorithm) for Task First strategy. There exist complicated relevancies among the earth observing sub-tasks in the united scheduling, thus makes the united scheduling a constraints optimization problem. Based on the immune genetic algorithm, the dissertation proposes a new algorithm named HIGA which adopts the permutation based coding and a two-level coding. By introducing a hierarchical control mechanism, HIGA uses the genetic operator in the upper level and the immunity operator in the low level. Moreover, in the immunity operator, the algorithm adopts gene recombination, adaptive immunity updating and the niche mechanism to improve the performance of the searching process. Experimental results show that the algorithm and the improved operators is both feasible and effective in solving the problem of multiple satellites scheduling with complicated tasks being in scheduling. (5) We propose the united scheduling process and algorithms for the task of moving target surveillance. Focusing on the task of moving target surveillance, the dissertation proposes a two-phase scheduling process including searching phase and surveillance phase. In the searching phase, in which condition there is no transcendental information, firstly, we uniformly partition the area using grid. Then we assume that the targets are randomly distributed in the searching area, and they are moving stochastically. Finally, we update the distribution probability of the moving target according to the historical observing results. Based on the above processes, a searching algorithm is proposed based on the dynamic updates of the distribution probability. In the surveillance phase, in which condition part of the transcendental information is observed, considering the effects of the geographical information and apriori information to the moving targets, an adaptive interactive multiple model algorithm for moving target surveillance is proposed, which supports the adaptive update and selection of the moving target movement model. Experimental results show that the proposed method is feasible and effective.