【作者】 史也；

【导师】 梁斌；

【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】 随着空间技术的发展以及空间开发与应用能力的不断提高，通过空间机器人进行自主在轨服务的意义和作用日益凸显，而空间机器人自主捕获目标的轨迹规划与控制是实现在轨服务的关键技术之一。本文结合国家863计划航天领域某项目，以对目标航天器自主捕获及捕获后目标停靠等在轨服务操作为应用背景，开展了空间机器人动力学建模仿真、目标航天器自主捕获、目标抓捕后复合体姿态调整和目标转移路径规划、目标捕获后轨迹跟踪控制、半物理仿真试验系统等方面的研究，以期推动我国空间机器人在轨服务技术的发展。针对空间机器人动力学建模进行研究，基于拉格朗日法推导空间机器人系统运动学和动力学方程，并对其非完整约束特性进行分析，以此作为研究非完整路径规划和轨迹跟踪控制的基础。此外，为满足空间机器人地面半物理仿真试验系统对实时性、计算精度以及稳定性的要求，引入R-W方法，从计算多体动力学的角度建立空间机器人动力学模型，并在此基础上实现了动力学计算软件，通过仿真验证了动力学模型的有效性。空间机器人自主捕获目标是在轨服务任务得以实施的前提和先决条件。本文研究了空间机器人自主目标捕获的视觉伺服轨迹规划方法，建立虚拟机械臂模型，并在此基础上采用蒙特卡罗法对空间机器人工作空间进行分析与仿真。为避免视觉伺服捕获过程中目标偏离相机视场、规划的关节角、角速度超出限位问题，提出了一种多约束条件下目标捕获的视觉伺服协调控制方法，包括相机视场约束和关节极限约束排斥力函数的建立，基于人工势能场法在图像空间生成规划轨迹，并通过图像雅克比矩阵将其映射为机械臂末端运动，同时引入协调控制方法减少机械臂运动所引起的基座姿态偏差；仿真结果表明该方法可实现在基座姿态、相机视场及关节运动范围受限条件下的目标捕获。针对空间机器人目标捕获后复合体姿态调整和目标转移路径规划问题，通过正弦多项式函数对机械臂关节角轨迹参数化，根据基座姿态和机械臂末端位姿控制精度指标设计目标函数，由此将自由漂浮空间机器人的非完整笛卡尔路径规划问题转换为非线性系统的优化问题；进而采用量子粒子群优化算法对该非线性优化问题进行求解，实现了非完整路径规划的目标，并通过仿真验证了该方法的有效性。针对空间机器人捕获目标后系统参数不确定情况下的轨迹跟踪控制问题进行了研究，推导了自由飞行空间机器人参数线性化形式的动力学模型，设计了反步自适应控制律，并给出了控制器参数与系统暂态性能之间的关系。针对自由漂浮空间机器人系统，设计了一种将约束输出转化为无约束控制变量的变换，然后针对此无约束控制变量设计了神经网络反步自适应控制律，仿真结果表明该方法可实现轨迹跟踪的稳定控制并满足一定的时域性能要求。为了降低技术研究与开发的风险和成本，确保在轨任务的成功，结合目前我国空间机器人系统研制任务对地面仿真试验的需求，设计并实现了一套与实际系统电气接口完全一致的空间机器人半物理仿真试验系统，并对本文所提出的方法进行了验证。更多还原

【Abstract】 The number of launched spacecraft has progressively increased year by year asthe ability to exploit and apply technology in outer space has developed, therefore,great advantages and enormous economic benefits can be gained by on-orbitservices through space robots. The key points of on-orbit services are the trajectoryplanning and control of autonomous capturing. This dissertation is prepared inaccordance with a project sponsored by the National High-Tech R&D Program ofChina (863Program). The research problem include modeling of the space robot,the autonomous capturing of target spacecraft, the attitude adjustment and pathplanning of the target after capturing, the trajectory tracking control after capturingand the development of semi-physical experiment. The main contributions of thisdissertation are as follows.Firstly, general kinematics equations and dynamic equations of space robot areestablished based on Lagrangian methods. The features of non-holonomic constraintare analyzed and utilized as the foundation of nonholonomic path planning andtrajectory tracking control design. In addition, in order to satisfy the requirements ofreal-time performance, calculation accuracy and stability for the space robot groundsemi-physical simulation system, the R-W method, proposed by Robertson andWitten Berg, is used to established the associated matrix and path matrices whichdescribed the topology relationship of the system. The relative motion equation ofthe adjacent rigid body are derived, and the dynamics equations of the space robotsystem are then established on the basis of the Virtual Power equation.Secondly, the visual servo method is studied for autonomous target capturingby a space robot. The workspace of the space robot is obtained by Monte-Carlomethod. Then, in order to avoid the possibility that the target moves out of thecamera’s FOV and the planned joint angle exceeds the limits, a visual servotrajectory planning method with multi-constraints is proposed. The camera FOVconstraint equation and the joint movement limits repulsive equation are established.The planned trajectory is generated in the image space on the basis of the artificialpotential field method, and then the trajectory in the image space is mapped to themovement of the end effector according to the image jacobian matrix. In order toreduce the disturbance aroused by the dynamic coupling of the manipulator and thebase, coordinated control is introduced into visual servo operation.Thirdly, path planning for the attitude adjustment of the composite body andthe target transfer are studied. A non-holonomic cartesian space path planningmethod based on quantum particle swarm optimization is proposed based on the nonholonomic redundancy features of a free-floating space robotic system. Thetrajectory of the manipulator joints is parameterized, and the target function isestablished by the control accuracy of the end effector and the base attitude. Thetrajectory-planning problem of a free-floating space robot converted into anoptimization problem of nonlinear system. The optimization problem of nonlinearsystem is then solved by quantum particle swarm algorithm. The solved parametersare substituted into the trajectory equation of the robotic arm to realized the goal ofnonholonomic trajectory planning. The effectiveness of the proposed method isverified by simulation.Fourthly, when the target is captured by a space robot, trajectory trackingcontrol problems with the uncertainty parameters of the kinematic and kineticequations are researched. The control law is designed based on back-stepping andadaptive control theory for free-flying space robot system. The relationship betweenthe control parameters and system transient performance is obtained. For a free-floating space robot system, in order to make the system’s tracking error meet theresponse constraints in time domain, one method that transforms the constrainedoutput to unconstrained control parameter is designed, then the back-steppingadaptive neural network control is designed to realize unconstrained variablestability control. Not only can the tracking error with global asymptotic stability ofthe system in the task space can be ensured according to this method, but also thetracking error of the closed loop system in a predetermined time domain response,which significantly improves the tracking performance of the system.Finally, in order to reduce the risk of research, ensure on-orbit mission success,and verify the method proposed in this paper, the real-time semi-physical simulationsystem whose electric interface is identical to a real system is established to meetthe mission requirements in space robot area.更多还原