【作者】 樊琼剑；

【导师】 沈春林；

【作者基本信息】 南京航空航天大学 ， 控制理论与控制工程， 2008， 博士

【摘要】 随着单架无人机技术的发展日趋成熟,军事和民事领域对无人机的任务需求变得苛刻,人们开始关注生物界编队鸟群(如大雁、天鹅等)长途迁徙的现象,分析生物系统的进化特征与行为规律,利用多无人机协同编队飞行(Coordinated Formation Flight,简称CFF)与生物系统(个体或群体)的某些原理和行为相似性,将仿生学引入到CFF研究中,以期获得类似鸟群长途迁徙的功效,如降低飞行阻力、节省燃油、延长巡航距离等。由于多无人机CFF控制技术具有广阔的工程应用前景,因此这一项目已在世界范围内激发了科研人员越来越高的研究热情,但又因该项目需要涉及多学科和多技术领域,因此研究难度高。目前国外虽已取得了显著的研究成果,但离工程应用还有很大的差距,而国内研究才刚刚起步,还属于理论跟踪性研究,所以系统深入的研究多无人机CFF控制技术,逐步实现其工程应用已成燃眉之际。本文正是基于多无人机CFF控制技术的国内外发展背景,根据实验室的实际情况,从多无人机编队飞行的基本原理到功能的硬件实现,采取环环相扣的研究方法,完成了多无人机CFF控制技术的前期研究工作。全文研究的多无人机CFF控制关键技术主要包括四个方面:多无人机CFF的气动耦合模型、CFF中单架UAV的运动学和动力学模型、CFF控制器以及硬件在环的CFF测试平台构建技术。论文首先总结了前人在这一领域内已有的研究成果,并在此基础上对紧密编队飞行中非常重要的气动耦合问题进行了系统的研究,然后分析对比了几种常见的涡流模型,利用简化的飞机结构和一种近似平均有效风和风梯度的计算方式,针对“长机-僚机”的V型编队方式和非线性6 DOF的刚性飞机,确立了适合多无人机CFF动态特性研究的气动耦合模型,继而分析这种气动耦合对飞机各种参量所产生的影响作用,并相应完成了对已有的标准飞机气动力和力矩系数方程组的调整工作。其次,利用第一阶段的工作成果,论文给出了“长机-僚机”编队方式下多无人机CFF模型,通过惯性坐标轴系、速度坐标轴系与机体坐标轴系之间的转换关系,深入的分析了受翼尖涡流影响的CFF中单架无人机的运动特性,同时给出了其特有的运动学和动力学模型。论文的核心研究内容之一是如何设计出一种能够确保僚机实时跟随长机飞行航迹的飞行控制器。在本文前期工作的基础上,利用多无人机CFF中的单架无人机的非线性动力学模型,针对飞机特有的运动规律,即飞机的状态变量可按时间尺度的不同分成慢变量(θ,ψ,?)和快变量( p, q ,r),对应的给出了双环控制器的设计方法:外环利用带积分消除跟随航迹稳态误差的变结构滑模控制器,内环则采用基于神经网络消除逆误差的动态逆控制器。整个设计过程紧紧围绕多无人机CFF系统建立的要求,由长机航迹信息已知的理想假设,到完全不用知晓情况下实施目标跟随,并保持特定的编队队形,层层深入地系统研究了飞行跟随控制律,最后利用Matlab7.1对其进行仿真验证。仿真结果表明该飞行控制器能够确保僚机在长机产生的涡流场中保持编队飞行的队形结构。本文另一个核心研究内容是硬件在环的多无人机CFF测试平台的研制。文中详细的阐述了多无人机CFF系统的设计要求和软硬件实现过程。整个系统主要由三个子系统组成:无人机飞行控制系统(Flight Control System,简称FCS)、基于Statemate构建的无人机虚拟样机(Virtual Prototype,简称VP)以及地面测试系统。硬件测试平台的设计中加入了FCS-VP思想,主要是基于低成本考虑,而FCS-VP虽然是一种数字化的软件模型,但其设计理念与系统设计自动化(System Design Automation,SDA)完全一致,可以对应的完成物理原型应该具备的所有功能,且具有研究过程用时短,飞行航迹监控实时性强等优势,并能随机的对飞机实施各种干扰,动态的显示编队飞行控制器的性能好坏。经过多次双机编队飞行的检测实验,结果表明基于多无人机CFF测试平台系统的双机编队飞行正常,达到设计要求,同时也进一步证明了本文所研究的编队飞行控制系统相关理论算法是正确和有效的。更多还原

【Abstract】 With maturing development of the single UAV technology, task requirments of UAV are changed to rigorousness in military and civil fields, and then more and more attention is paid to the flight phenomena made by migration birds (such as Canadian geese, swan) during long distance journey, which suggested followers are able to take advantage of significant amount of power-savings due to drag reduction induced by the vortices generated from the leader’s wings. In the wake of this feature of migration birds, more and more study passions are focused on the project of multi-UAV biomimetic CFF in the whole world in order to accomplish the complicated missions safely and effectively in military field. As great benefits are drawn from the applications of multi-UAV CFF, especially the applications, so it is urgent to enhance the research work of this field to shorten the gap to the foreign countries and accumulate knowledge in theory aspect for the development of our UAVs in the future. This thesis analyzes multi-UAV CFF from four major standpoints: aerodynamic coupling, dynamic modeling, control desiging and CFF testbed based on hardware-in-the-loop.This thesis discusses firstly the very important problem of aerodynamic coupling involved in close-proximate flight in-depth after summarizing previous researches on multi-UAV CFF, and introduces three different vortex models, and then a simple and versatile multi-UAV model has been evolved by introducing“Leader- Follower”frame as its reference in the dynamic modeling. Nonlinear 6-DOF, rigid body, equations of motion developed in the LF frame are used to model the UAVs in the group. The nonlinear equations of motion contain the wind effect terms and their time derivatives to represent the aerodynamic coupling involved in CFF. These wind terms are obtained using an averaging technique that computes the effective induced wind components and wind gradients in the UAV’s body frame. This way, the equations of motion based on the assumption that the plane is exposed to uniform wind components and gradients can be used. This is because the non-uniform wind field is very complicated to analyze multi-UAV CFF dynamic characteristics so that the averaing technique is used to approximate the non-uniform wind field as uniform wind components and gradients.An algorithm that generates a safe and feasible trajectory for the formation-keeping and formation-reconfiguration of multi-UAV CFF is one of kernel study contents of this thesis. It is a combination of integral controller, including Variable Structure System Sliding Mode Control and Nonlinear Dynamic Inversion Control with Neutral Network. A nonlinear, 6 DOF multi-UAV simulation has been developed in MATLAB 7.1 to test the efficacies of the dynamic models and the nonlinear controller. Simulation results show that the controller is capable of producing both efficient formation-keeping and formation-reconfiguration under the effect of vortex-induced wind.At the end of this thesis, the other kernel study content is the design and building process of hardware-in-the-loop testbed system of multi-UAV CFF which is described in detail. Three subsystems constitute the testbed system: the flight control system based on DSP technique, the flight control system-virtual prototype based on Statemate technique and ground test subsystem. Although, Virtual Prototype (VP) is digitized software model, it is a new technology in correspondence with the idea of System Design Automation (SDA), which realizes all functions of physical prototype. Finally, we implement the proposed control scheme on the testbed, and the flight tests display successfully that the follower can fly along the flight path of the leader UAV and keep formation distance, and to further ensure that the theory and design of multi-UAV CFF controller are effective and correct.更多还原