【作者】 杨成顺；

【导师】 杨忠；

【作者基本信息】 南京航空航天大学 ， 导航、制导与控制， 2013， 博士

【摘要】 多旋翼飞行器简单的结构、超强的机动性、独特的飞行方式以及在军事和民事领域展现出的巨大应用价值，引起了国内外学者以及科研机构的广泛关注，并迅速成为目前国际上研究的热点之一。然而，国内对多旋翼飞行器自主飞行控制的研究与国外先进水平之间还有一定的差距，此外，目前所广泛研究的四旋翼飞行器的旋翼组件无任何冗余，任一执行机构发故障，都难以实施容错飞行控制，只能任其坠落。因此，多旋翼飞行器在构型以及先进飞行控制方法方面都需要进一步深入研究。针对上述问题，本文以课题组提出的一种面对称布局多旋翼飞行器为研究对象，以其实际飞行需求为目标，分别在数学建模和先进飞行控制等方面开展了一系列研究工作。首先，在详细分析飞行器的构型以及飞行原理的基础上，对旋翼以及整个机体的空气动力和动力矩进行分析计算；在此基础上，详细推导并建立出较为完整的飞行器六自由度非线性数学模型以及飞行运动控制模型，同时对数学模型和控制模型的开环特性进行了检验和分析，论证了该飞行器的飞行运动特性具有强非线性、强耦合、不确定的特点，可以满足飞控系统设计的需求。而后，在充分考虑飞行器的不确定性以及外部干扰因素下，推导出带有复合干扰的姿态控制系统模型。针对不确定姿态系统，提出一种基于全调节径向基神经网络干扰补偿的鲁棒自适应姿态控制器，其中神经网络的中心向量和宽度向量均可在线进行调节。仿真结果表明该方法可有效实现对干扰的抑制，并可满足飞行器的姿态运动控制要求。接着，为实现飞行器精确跟踪期望轨迹的任务需求，提出一种基于回馈递推-终端滑模变结构控制的轨迹跟踪控制方法。其中，为抑制不确定以及外界干扰等构成的复合干扰对飞行器的影响，提出一种自适应滑模干扰重构算法，该算法和径向基神经网络一样都无需知道复合干扰的界，但所提算法与径向基神经网络相比具有设计简单，易于工程实现的优点；为实现该飞行器姿态稳定以及对期望姿态和轨迹的快速与精确跟踪，将所提的自适应滑模干扰重构算法与回馈递推控制方法、动态面控制方法以及终端滑模变结构控制方法相结合设计了飞行器的姿态控制器并基于回馈递推控制方法设计了飞行器的位置控制器。仿真结果表明所提方法的有效性和正确性。然后，为了更好地体现和满足飞行器实际工作特点和飞行需求，进一步综合考虑该飞行器在复合干扰影响以及状态受限情况下的轨迹跟踪控制问题。其中，针对系统中的复合干扰，设计了滑模干扰观测器用于对复合干扰的重构和补偿控制律的设计；针对该飞行器位置控制模型的特殊形式，借鉴滑模控制方法的设计思想设计了飞行器的位置控制器；在姿态控制器设计中考虑复合干扰的影响、姿态角以及角速率受限要求，设计了基于复合干扰重构补偿以及受限指令滤波的回馈递推姿态控制器。结果表明所提方法可实现该飞行器在复合干扰以及状态受到约束情况下的轨迹跟踪控制。最后，针对飞行器姿态控制系统的执行器故障问题，提出一种故障诊断、辨识和容错控制策略。其中，考虑文献[199]滑模干扰观测器的控制量中存在着切换函数，可能导致复合干扰的重构值存在抖振现象的问题，提出一种改进滑模干扰观测器复合干扰重构算法，并针对执行器的失效和卡死故障问题，提出一种类MIT模型参考自适应控制的局部执行器故障估计算法；基于所提的改进滑模观测器复合干扰重构算法和执行器故障估计算法与回馈递推控制方法结合设计了该飞行器姿态系统的容错控制器。通过Lyapunov方法进行了闭环性能分析，同时仿真得到了满意的控制效果。更多还原

【Abstract】 Due to its simple structure, superior maneuverability and unique flying way, multi-rotor aircrafthas attracted increasing attention from scholars and research institutions at home and abroad and it isbecoming one of the attractive research fields. However, the greater disparity exists when comparingthe autonomous flight control of the multi-rotor aircraft domestic with that of abroad. Besides, it isdifficult to implement fault-tolerant flight control when any of the actuator breaks down because ofthe rotor assembly of the quad-rotor without any redundancy. Therefore, the configuration and theadvanced flight control method of the multi-rotor aircraft is to be studied furtherly. Accordingly, thedissertation carries out a series of research work in modelling and advanced flight control to meet thetarget of the actual flight for a novel multi-rotor aircraft which was awarded a national inventionpatents.First of all, the aerodynamics forces, torques of the rotor and the whole airframe are analyzedand derived on the basis of the detailed analysis of the configuration and the principles of flight of theplane symmetrical multi-rotor aircraft. Then, the systemic nonlinear mathematical model with sixdegree of freedom and the flight control model are developed for the aircraft, respectively. At thesame time, the open-loop characteristic of the mathematical model and control model are proved andanalyzed. They demonstrate that the developed model can embody the characteristics of the aircraft,such as serious nonlinearity, intense state-coupling and uncertainties and can meet the requirements offlight control system design.Secondly, an attitude control system model with composite disturbance of the aircraft is deducedunder taking full account of the uncertainties and external disturbance factors for the aircraft. And, arobust adaptive attitude controller with compensation the disturbance based on fully tuned radial basisfunction neural network is proposed for the uncertain attitude system and the center vector and thewidth vector of the neural network all can be adjusted online. Simulation results show that theproposed method can restrain the disturbance effectively and can meet the requirement of the attitudemotion control.Thirdly, a method of trajectory tracking control for the aircraft is proposed based onbackstepping terminal sliding mode control to achieve the requirement of tracking the desiredtrajectory accurately. Thereinto, an adaptive sliding mode algorithm for disturbance reconstruction isdeveloped to restrain the impact of the composite disturbance on the aircraft. The algorithm does notneed the boundary of the composite disturbance just like the RBF neural network. But, compared with RBF neural network, the proposed algorithm is simple and fit for engineering. In order to achieve theattitude stabilization and to track the desired attitude and trajectory quickly as well as accurately, anattitude controller is designed which use a combined adaptive sliding mode algorithm for disturbancereconstruction, backstepping, dynamic surface control and terminal sliding mode control. After that, aposition controller is designed based on backstepping. The effectiveness and correctness of theproposed methods are verified through the simulation results.In the following, a problem of trajectory tracking is further considered under the condition ofcomposite disturbance impact, the state constraints to better reflect and meet both the actual flightcharacteristics and requirements. Aiming at the composite disturbance of the system, a sliding modedisturbance observer is designed to reconstruct the composite disturbance and to bulid thecompensatory control law. Considering the special form of the position control model, a positioncontroller is referred from the designing ideas of the sliding mode control and developed. The impactof the composite disturbance and the requirements of the attitude angles and angular velocitiesconstraints are considered in the attitude controller design. Then, a backstepping attitude controller isdesigned based on composite disturbance reconstruction and compensation, as well as constrainedcommand filter. The simulation results demonstrate that the proposed methods can make the aircrafttracking the desired trajectory under the case of composite disturbance and the state constraints.Finally, aiming at the problem of actuator fault of the attitude control system, a novel faultdiagnosis, identification and fault tolerant control strategy is presented. In ref.[199], there is aswitching function in the control variable of the sliding mode disturbance observer which may lead tothe problem of chattering phenomena of the composite disturbance reconstruction value. A modifiedsliding mode disturbance observer is proposed to reconstruct the composite disturbance. Consideringthe actuator fault models include both the loss of actuator effectiveness fault and the actuator-stuckfault, a local actuator fault estimation algorithm is presented which is similar to the mode referenceadaptive control. Then, an attitude controller based on the proposed composite disturbancereconstruction algorithm, the actuator fault estimation algorithm and backstepping is designed. Thestability of closed-loop system is analysed via Lyapunov method and the excellent performance isdemonstrated by simulation.更多还原