【作者】 袁国平；

【导师】 史小平；

【作者基本信息】 哈尔滨工业大学 ， 控制科学与工程， 2013， 博士

【摘要】 伴随着空间技术的发展，航天器的姿态控制一直是航天器技术研究领域的热点和难点问题之一。目前，许多的空间任务（例如对地观测、编队飞行等）要求航天器具有良好的姿态控制性能。航天器迅速、精确的完成空间任务可以增加其使用的范围，获得更多有价值的信息。但航天器的姿态跟踪或姿态机动是一个典型的非线性控制问题，控制器综合难度大。而且现代航天器上挠性结构的增多也导致振动问题严重。另外，航天器在轨运行期间，不可避免的会受到各种环境力矩的干扰；同时，航天器执行部件安装误差等因素造成输出力矩的偏差也会影响姿态控制精度。传统的航天器姿态控制方式已逐渐不能适应许多新型空间任务对控制性能的需求。随着近年来控制理论的发展，许多非线性控制方法被用于设计航天器姿态控制器，在这种背景下，从理论研究和工程实践的角度出发，本文对航天器的姿态控制，挠性结构的主动振动抑制和太阳帆板驱动时的姿态补偿控制等问题进行了深入的探讨。本学位论文主要研究内容包含以下四个方面：（1）建立了航天器的运动学和动力学模型，然后着重给出太阳帆板驱动机构与航天器姿态的耦合模型，作为后续控制系统的分析与设计的基础。随后，以挠性航天器的平面姿态机动的建模与控制为例，提出一种符号计算在航天器仿真中的应用方案，为文中仿真平台设计与开发提供必要的技术支撑。（2）针对带有转动惯量矩阵不确定性、受外干扰作用以及飞轮作为执行器存在安装误差的航天器姿态控制问题，提出一种基于自适应鲁棒方法思想的控制律，并给出了严格的稳定性证明，该方法以反馈线性化方法为基础并采用径向基神经网络（Radial Basis Function Neural Network，RBFNN）设计补偿策略。随后，考虑执行器在实际工作中存在饱和的现象，针对一类带有不确定性的非线性Hamilton系统的跟踪控制问题，提出了一种具有抗饱和能力的自适应鲁棒控制律，该策略同样以反馈线性化方法为基础，通过构建扩张状态观测器（Extended State Observer,ESO）得到外干扰的量测值并将其引入控制律中补偿干扰带来的性能损失，同时在控制律中引入一个辅助系统，并对其动力学进行设计，可减小控制饱和对系统性能影响，并基于Lyapunov理论分析了闭环系统的稳定性。上述控制方案被用于航天器的姿态跟踪控制中，通过数值仿真验证了该方法的有效性。（3）针对带有不确定性的挠性航天器的姿态控制问题，提出了一种分散自适应鲁棒姿态控制算法，保证航天器受外界干扰作用和存在参数不确定性情况下，姿态能精确的跟踪指令信号的变化。设计中，先将挠性航天器模型变换成三个子回路分别进行控制器综合，利用扩张状态观测器获取子回路间的耦合特性及外干扰的量测值进行补偿后实现子回路间的解耦，然后设计自适应律对模型中不确定参数进行辨识，最后设计鲁棒控制项保证整个系统的稳定性。考虑到上述方法的不足，为兼顾挠性航天器的姿态控制精度和对挠性振动的抑制效果，提出了一种姿态控制与振动抑制相结合的复合控制方法。采用LQR方法设计了挠性附件的主动振动控制器，保证挠性振动的快速衰减，给出并分析了一种Q,R矩阵的选择方法；在姿态控制器设计时，反馈线性化理论仍是整个方法的基础，不确定性参数的补偿策略采用动态模糊神经网络（Dynamic FuzzyNeural Network，D-FNN），D-FNN可根据系统的性能动态调整网络的结构和规则的条数，有效避免传统神经网络结构和参数选择的盲目性，然后分析挠性附件的振动及外干扰对姿态影响的界函数并将其作为鲁棒控制项设计的基础加入到姿态控制律中以保证整个系统的性能。（4）考虑挠性航天器对地定向任务中太阳帆板驱动对姿态控制性能产生影响的问题，提出一种姿态与驱动机构的复合控制方法。为避免由太阳帆板驱动机构开环控制带来的转速波动问题，以坐标变化和反馈线性化方法为基础，提出了一种角速率闭环的鲁棒控制律，利用扩张状态观测器设计补偿律抵消波动力矩的影响达到提高驱动速率平稳性的目的。在此基础上，考虑对地定向任务中的姿态稳定控制，设计了前馈补偿算法降低帆板驱动机构对姿态控制性能的影响，保证了高精度的姿态控制。随后，数值仿真对算法的有效性进行了验证。考虑到本文研究中仿真程序开发的复杂性，为降低设计难度并提高开发效率，对基于模块化、易扩展和面向航天器控制仿真平台的开发方案进行了研究。从任务背景出发提炼平台的功能需求并对基于分层设计的仿真任务软件的体系结构进行了阐述。从控制系统设计的角度，综合利用Matlab/Visual C++等软件，对航天器导航、制导与控制系统的部件进行模型构建。为保证仿真平台能够根据不同的仿真任务灵活的配置部件模型及相应的参数，设计了基于用户接口界面的模型管理系统，完成整个仿真平台的设计与开发。更多还原

【Abstract】 With the fast development of aerospace technology, spacecraft attitude controlproblem has always been a hot and difficult issue in this field. Now, many aerospacemissions, such as earth observation, formation flying and so on, often requireachievement of the desired states with good control performance. Spacecraft withthe ability of completion of aerospace mission rapidly with high precision canincrease its usage ream to get more valuable information. Spacecraft attitudetracking or maneuver is a typical nonlinear control problem and the controllersynthesis is difficult. The increase of the flexible structure in modern spacecraft alsoled to serious vibration problems. In addition, the spacecraft in orbit is affected byvarious environmental disturbance torques; at the same time, the deviations of thecontrol torques caused by the implementation of parts of the spacecraft installationerror also affect the accuracy of attitude control. The traditional spacecraft attitudecontrol method has not been gradually adapted to the many new aerospace missions.With the development of control theory, many nonlinear control schemes are used tosolve the problems of spacecraft attitude control. Under this background, thisdissertation focuses on spacecraft attitude control, flexible vibration suppression andattitude control considering a solar array. The main researches are given as follows:(1) As the foundation of the controller analysis and synthesis in latter chapters,the kinematic equation and dynamic equation are formulated, and the mutualcoupling between the spacecraft dynamic and the solar array drive assembly areespecially given. Then, the model and control of the flexible spacecraft maneuverused as an example, a frame of spacecraft simulation with symbol calculation isproposed in this chapter which is the basis of the simulation platform design anddevelopment latter.(2) Considering the attitude tracking control problem of the rigid spacecraftactuated by fly-wheels, a control scheme based on adaptive robust method ispresented and the stability is proved strictly. Feedback linearization method is thefoundation of the proposed control method and radial basis function neural networkis introduced to design compensation law. In the following part, considering theactuator outputs are limited in actual applications, for the tracking problem of aclass of nonlinear Hamilton system with control saturation, a anti-windup robustadaptive control law is studied based on feedback linearization theory, whereextended state observer is constructed to estimate and reduce the influence ofexternal disturbances and san auxiliary signal is introduced into the control law to compensate for the effect caused by control saturation, then the stability of theclosed loop system is analyzed based on Lyapunov theory. Based on the methodabove, a spacecraft attitude controller considering saturation is designed. Thefollowing numerical simulation is given to illustrate the efectiveness of thedesigned controller.(3) A robust adaptive decentralized control algorithm for a flexible spacecraft isproposed in the presence of external disturbances and parametric uncertainties,which can guarantee the attitude track the command signal accurately. In the designprocedure, the model of the flexible spacecraft is divided into three sub-loops forcontroller synthesis separately and the extended state observer is introduced to getand compensate the nonlinear coupling among the sub-loops and externaldisturbances, then an adaptive law is designed to estimate the uncertain parameters,at last the robust control part is used to achieve the stability of the closed loopsystem. Considering the drawback of the above method, for high attitude controlaccuracy and good flexible appendage vibration suppression effect, the LQR methodis used to design the active vibration controller to ensure the fast decay of theflexible vibration. In order to reduce the blindness of parameter selection, a Q, Rmatrix selection method is given; in the attitude controller design procedure, theproposed method is still based on feedback linearization theory and the dynamicfuzzy neural network (D-FNN) is used to design the compensation scheme for theuncertain parameters. The structures and rules of the D-FNN can be dynamicallyadjusted according to the performance, which can effectively avoid the irrationalityof the selection of the traditional neural network structure and parameters, after that,the robust control part based on the bounded function, which is obtained from theflexible appendage vibration and external disturbances analysis, is added to theattitude control law to ensure the performance of the entire system.(4) For improving the performance of flexible spacecraft attitude control with thespeed fluctuation of the solar array drive assembly (SADA), composite control ofthe spacecraft attitude and SADA is presented for the earth-orienting mission.Taking into account most of the SADA using open-loop control lead to speedfluctuation, an angular rate closed-loop robust control law is developed based oncoordinate transformation and feedback linearization method and the extended stateobserver is added to the control law to reduce the disturbance torque for improvingthe drive rate stability. On this basis, for the attitude regulation problem in theearth-orienting mission, a forward compensation algorithm is given to reduce theimpact of the SADA, and to ensure a high-precision attitude control. Theeffectiveness of the algorithm above has been verified via numerical simulation.Considering the complexity of the simulation program developed, a modular, easily expandable and spacecraft control simulation platform is proposed to reduce thedifficulty of design and improve development efficiency. The platform functionalrequirements are refined from the mission background and the simulation tasksoftware architecture is described based on the hierarchical design. From the view ofcontrol system design, the models of the components of spacecraft navigation,guidance and control are built on MATLAB/Visual C++software. In order toensure that the models of the components and the simulation parameters can beflexible configuration according to the different simulation missions modelmanagement system based on the user interface is designed, the entire simulationplatform is completed.更多还原