大规模对地观测卫星组网统筹运控理论与方法

Earth observation satellite technology, including agile satellite and mini satellite technology, is becoming mature. The type and quantity of satellite are gradually expanding, as a consequence, it has been gradually own the hardware foundation to construct large-scale satellite network in terms of reality meaning. Then it is also becoming an important research topic to explore an approach of co-ordinate operation control. In one side, it is not conducive to the overall control when satellite resources are fully decentralized in some important departments, in the other side, it is lack of fine management flexibility in terms of important departments when satellite resources are fully centralized into one department. Therefore, a new kind of management mode, namely group control mode, is proposed in this project for large scale satellite network. Five respects of work are done to group control model. Firstly, group control mode is proposed, which tries to make balance between two respects of consideration: both the overall coordination control and the management flexibility by each important department. In the project, Multi-Agent is applied to form the entire control system architecture. Secondly, various service requirements are well-designed and are fully blended, before establishing a collaborative group planning model. Two kinds of model include static planning model and interactive dynamic planning model. There are several satellite types of large-scale satellite network, including optical imaging satellites, electronic reconnaissance satellites, mapping satellite etc. Collaborative optimization models are constructed, which not only deals with large-scale simply missions but also deals with the complicated collaborative missions. Thirdly, data mining technologies are adopted to collect mission collaborative modes which are helpful to describe the complicated mission, and are also easily represented by the collaborative relation. With introducing the corresponding collaborative relation, the collaborative work among different satellite is arranged by the planning model. Fourthly, the intelligent collaborative optimization algorithm for large-scale solving is intensive explored from view of the field of heuristic information mining, optimal combined algorithm, distributed parallel processing, adaptive intelligent collaborative work. Fifthly, Bayesian network is applied to build risk assessment model, and robust factor is introduced so as to improving the performance of robust for overall coordination control.

随着对地观测卫星技术发展，包括敏捷卫星、微小卫星新一代技术也正日趋成熟，卫星类型和数量正逐步扩大，大规模组网已逐步具备现实的硬件基础，探索国家层面上如何统筹运控成为一项重要研究课题。资源完全分散在多个重要部门不利于统筹运控管理，而资源完全集中又缺少各部门精细化管理的灵活性。为此，本项目面向大规模卫星组网提出群运控模式，兼顾精细化管理与统筹运控需求，实现带有多个重要部门协调的统筹运控管理；本项目考虑多种业务融合下的运控需求，建立协同式群规划模型，实现包括光学成像卫星、电子侦察卫星、测绘卫星等多类型大规模卫星组网的群规划技术；提出智能协同优化求解方法，从领域启发信息挖掘、组合算法优化、分布并行式处理、自适应智能协同工作等角度探索实际大规模运控系统求解技术；应用贝叶斯网络进行风险评估，探寻风险控制手段，将鲁棒性要素纳入统筹运控中，提高群运控风险应对的鲁棒性。

对地观测卫星可为人们提供重要作用，在如军事冲突、反恐、农业、灾害、测绘、气象、资源环境、公共安全等重大问题中，卫星具有侦查面积大、范围广、速度快、可长期或连续监视及不受地理条件限制等优点。在航天大国中，若干重要部门都已有自己专属的卫星资源，一方面，可以依据本门自身的业务需求进行运行规划，实现按部门精细管理，另一方面，各部门受资源类型和数量限制，观测能力有限，各部门之间有效协作会获得更大效益。本项目站在国家层面上，提出群运控模式进行大规模组网，实现联合各重要部门的统筹运控管理，是一种兼顾部门精细管理和全局统筹运控的管理模式。.本项目主要研究有三方向：（1）群运控模式研究，该方向包括：卫星管理组织与群用户协商机制、Multi-Agent架构设计与资源的Agent建模、统筹运控技术研究；（2）协同式群规划建模研究，该方向包括：协同模式挖掘与协同关系表示、群任务规划调度建模研究、群运控风险控制；（3）智能协同优化求解研究，该方向包括：大规模优化求解技术研究、智能协同优化求解、Multi-Agent统筹运控的合同网求解技术。.对上述三个主要方向开展研究，本项目达到以下四项目标：（1）构建大规模卫星组网的群运控模式。面向大规模卫星组网提出群运控模式，兼顾精细化管理与统筹运控需求，实现带有多个重要部门协调的统筹运控管理，又对该运控模式的各个要素进一步界定，丰富了群运控模式的内涵和外延。（2）探索群运控理论方法。经过对卫星运控中各参与元素，利用Agent建模，构建Multi-Agent系统，实现组网卫星系统中的资源、用户、任务、管理之间相互协调。（3）建立多星协同优化调度的模型。对各类复杂协同任务进行统一描述，并在任务规划约束满足建模中体现协同任务约束。（4）实现具有自适应能力的群运控大规模求解方案。充分利用群运控、启发式求解、分布式求解等特点提高统筹规划求解的收敛性和稳定性。