毕业生知识库

近几十年来，无人机在军事和民用领域得到了广泛的应用，并展现出广阔的应用前景。而随着无人机面临的任务更加复杂化和多样化，相较于单架无人机，多无人机协同编队飞行由于其优越的灵活性和感知能力成为了航空领域的研究热点。作为多无人机编队设计中诸多关键技术之一，编队控制技术受到了控制领域科研人员的不断关注。近年来，新兴的多智能体一致性理论由于能通过智能体间信息交互与相互协调来实现各智能体的状态达到协同一致的优点，为多无人机的编队控制系统设计提供了新思路，在解决编队控制问题中发挥出独特的优势。然而，基于一致性理论的协同编队控制技术仍然处于探索和发展之中，诸多关键控制问题仍亟待解决。本文致力于实现多无人机时变队形编队跟踪飞行的控制目标，重点解决编队飞行中存在的收敛速度慢、时变时延、通信拓扑切换和不确定性等具有挑战性的控制问题。论文的主要工作和创新点归纳如下：
（1）为了实现多无人机系统的快速时变队形编队跟踪飞行，同时考虑多无人机系统在实际工作中存在通信拓扑变换问题，提出一种新型的基于齐次性理论的有限时间编队控制算法。该控制算法能够使多无人机系统在通信拓扑切换情况下，在有限时间内实现时变队形编队跟踪飞行。通过Lyapunov 理论证明了整个闭环编队控制系统的有限时间稳定性，同时通过仿真实验验证了提出的控制算法的有效性。
（2） 针对多无人机系统在实际工作中存在的时变时延和通信拓扑切换问题，设计基于一致性理论的集成式控制律。该控制律能够同时处理时变时延和通信拓扑切换问题，并且实现多无人机系统时变队形编队跟踪飞行。通过Lyapunov 理论证明了编队跟踪误差的渐近稳定性，同时通过仿真实验验证了提出的控制律的有效性。
（3） 同时考虑多无人机系统在实际工作中存在的不确定性和时变时延问题，提出一种新型的基于神经网络的分布式自适应编队跟踪控制框架。该框架能同时补偿系统不确定性和处理时变时延问题，并实现时变队形编队跟踪飞行。通过Lyapunov 理论证明了编队跟踪误差的最终一致有界性，同时通过对比仿真实验验证了提出的控制框架的有效性和优越性。
（4）针对系统不确定性问题和传统神经网络自适应控制器中自适应参数过多的问题，设计一种新型的基于最小参数学习（Minimal Learning Parameter，MLP）的编队控制律，使多无人机系统实现时变队形编队跟踪飞行的同时减少自适应律的计算量。通过Lyapunov 理论证明了编队跟踪误差的最终一致有界性，同时通过对比仿真实验验证了提出的自适应控制算法的有效性和优越性。
（5）围绕多无人机系统在编队飞行中存在的领航者速度信息不可得或者测量不精确的问题以及多无人机系统在实际工作中存在的不确定性问题，提出一种新型的基于最小参数学习和固定时间级联式领航者状态观测器（Cascaded Leader State Observer，CLSO）的自适应神经网络控制框架。首先，考虑观测器收敛速度对控制性能的影响，提出一个新型的固定时间收敛的CLSO，使得系统中每一个跟随者能够在不依赖领航者速度测量值的情况下，在固定时间内获得领航者状态的准确估计值。其次，针对系统不确定性问题，采用自适应神经网络进行在线补偿。最后，结合所设计的固定时间CLSO 和MLP 方法提出新型的编队飞行控制框架，在实现时变队形编队跟踪飞行的同时缓解系统的计算载荷。通过Lyapunov 理论证明了编队跟踪误差的最终一致有界性，同时通过对比仿真实验验证了提出的控制框架的有效性和优越性。
总体而言，本文从时变队形编队跟踪飞行的控制目标出发，深入研究了收敛速度慢、时变时延、通信拓扑切换和不确定性等若干关键控制问题，为无人机协同编队控制技术的发展做出了积极的理论探讨。

In recent decades, unmanned aerial vehicles (UAVs) are widely used in military and civil fields, and present vast and applicable prospects. With the increasingly complex and diverse tasks the UAVs face, comparing to the single-UAV flight, UAV cooperative formation flight has become a research hotspot in the aerospace field due to its overwhelming flexibility and sensing ability. As one of the significant technologies in UAV formation system design, UAV formation control technology has received constant attention from the researchers in the control field. In recent years, the emerging multi-agent consensus theory, which can lead the states of the agents to a common value by information interaction and coordination, provides a new idea for UAV formation control design and possesses distinct advantages in solving formation control problems. However, cooperative formation control based on consensus theory is still under exploration and development, and many key control issues need to be addressed. Aiming to realize the time-varying formation tracking flight of multi-UAV systems, this dissertation focuses on solving the challenging control problems, such as insufficiently fast convergence speed, time-varying delay, switching communication topologies and uncertainties. The main work and novelties of this dissertation are summarized as follows:
(1) To achieve fast time-varying formation tracking flight for multi-UAV system, a novel finite-time formation control law based on homogeneity theory is proposed with consideration of the switching topologies problem in real flight. Such a formation control law can achieve the target of time-varying formation tracking flight in finite time under switching topologies. The finite-time stability of the whole closed-loop formation control system under the control law is obtained based on Lyapunov theory. Numerical simulations verify the effectiveness of the proposed control strategies.
(2) For the problems of time-varying delays and switching topologies, a novel integrated consensus based formation control law is designed. Such a formation control law can simultaneously solve the problems of time-varying delays and switching topologies, at the meantime, realize the desired time-varying formation tracking flight. The uniform ultimate boundedness of the formation tracking errors is theoretically analyzed through Lyapunov approach. In the meanwhile, simulation results demonstrate the effectiveness of the control law.
(3) Considering the problems of time-varying delays and uncertainties in practice simultaneously, a novel radial basis function neural network (RBFNN)-based fully distributed adaptive control scheme is proposed. Such a control scheme can achieve time-varying formation tracking flight, at the same time simultaneously compensate for uncertainties and tackle the problem of varying time delays. The uniform ultimate boundedness of the formation tracking errors is obtained through Lyapunov approach. In the meanwhile, the comparative numerical simulation results verify the effectiveness and superiority of our proposed control scheme.
(4) Aiming at solving the problems of uncertainties and too many parameters needing to be updated online in the conventional RBFNNs, a novel formation control law based on MLP (Minimal Learning Parameter) is designed to achieve time-varying formation tracking flight for multi-UAV systems and lighten the burdensome computation of the adaptive control law at the same time. The uniform ultimate boundedness of the formation tracking errors is obtained through Lyapunov approach. In the meanwhile, the comparative numerical simulation results verify the effectiveness and superiority of the control law.
(5) Around the problems of uncertainties in practical flight and unavailable precise velocity states of the leader, a novel adaptive neural network formation tracking control scheme is proposed based on MLP technique and fixed-time cascaded leader state observer (CLSO) without velocity measurements. Firstly, taking the convergence rates of the observers into consideration, a novel fixed-time CLSO without velocity measurements is designed for each UAV to obtain the precise estimates of the states of the leader. Secondly, focusing on the problem of model uncertainties, adaptive RBFNNs are adopted to compensate for the model uncertainties online. Finally, a novel formation control scheme is proposed by combining the designed fixed-time CLSO and MLP method, to achieve formation tracking flight and lighten the burdensome computation of the whole system. The uniform ultimate boundedness property of the formation tracking error is obtained through Lyapunov approach. In the meanwhile, the comparative numerical simulation results verify the effectiveness and superiority of the control scheme.
On the whole, starting from the control objective of time-varying formation tracking flight, this dissertation deeply studies several key control problems, including insufficiently fast convergence speed, time-varying delay, switching communication topologies and uncertainties, and makes a positive theoretical discussion for the development of cooperative formation control of multi-UAV systems.