【作者】 廖一寰；

【导师】 唐国金；

【作者基本信息】 国防科学技术大学 ， 航空宇航科学与技术， 2011， 博士

【摘要】 机械臂在空间组装、维修等任务中有着广阔的应用前景，也是我国空间站建设的关键技术之一。本文以解决空间柔性机械臂的运动规划问题为目标，系统研究了机械臂的载体姿态最小扰动规划方法、减振运动规划方法和多目标最优的运动规划方法。主要研究成果如下：建立了空间自由漂浮柔性机械臂系统的动力学模型。利用ADAMS验证了动力学模型的正确性，并分析了动力学建模过程中模态截断引起的误差，以及载体自由漂浮状态对机械臂弹性变形的影响，为机械臂系统的运动规划研究奠定了基础。改进了点对点操作任务中对载体姿态无扰的运动规划方法。将机械臂的无扰运动规划问题离散化为参数优化问题，并结合遗传算法的全局搜索能力和序列二次规划算法的快速收敛能力，提出了一类含初值估计的串行优化策略。分别求解了平面运动及三维运动机械臂的无扰运动规划问题，对串行优化策略的全局收敛能力及有效性进行了充分的测试；提出了一类基于平推无扰曲线的运动规划方法，可有效改善采用参数优化方法求解无扰运动规划问题的收敛性；提出了基于无扰运动轨迹的速度规划方法，在考虑关节驱动力矩约束的条件下，得到了沿无扰曲线运动的速度以及最短操作时间。提出了点对点操任务中减振运动规划问题的混合规划策略。根据减振运动规划问题的特点，结合Gauss伪谱法的计算效率以及直接打靶法的精确性，提出了一类混合规划策略；分别求解了全程振动最小化与残余振动最小化的运动规划问题，验证了混合规划策略的有效性与鲁棒性；考虑到实际操作中机械臂初始状态的不确定性，建立了考虑残余振动鲁棒评价指标的运动规划模型，得到了鲁棒评价指标最小的关节运动。研究了连续路径跟踪任务的运动规划方法。提出了基于自运动的载体姿态最小扰动规划方法，将其与基于广义雅可比矩阵伪逆的运动规划方法进行对比，验证了该方法的适用性；将基于自运动的规划方法与基于Gauss伪谱法和直接打靶法的混合规划策略相结合，快速稳定地获得了连续路径跟踪任务中减振运动规划问题的解。发展了空间机械臂系统的多目标最优运动规划方法。提出在机械臂运动规划过程中，同时考虑载体姿态扰动、弹性振动与操作时间等设计指标；考虑到多目标最优运动规划问题的特点，利用物理规划方法将多目标最优问题处理为一个反映设计者偏好的单目标最优问题，并采用混合规划策略进行求解；利用NSGA-ΙΙ验证了物理规划解的Pareto最优性，然后求解了各种操作任务的多目标最优运动规划问题，均得到了满足设计者偏好的解。论文提出了空间机械臂的载体姿态最小扰动和减振运动规划方法，并探索了多目标最优运动规划的模型和算法，在理论和方法上都取得了一定的进展。论文研究紧密结合空间机械臂的应用背景，提出的各种运动规划方法对发展我国的空间机械臂具有重要的应用价值。更多还原

【Abstract】 The space manipulator which has a broad application in space tasks such as spaceassembly and repair, is one of the critical technologies for the foundation of our spacestation. Aiming to solve the motion planning problems of space flexible manipulator,this dissertation studies the approaches of motion planning for minimum disturbance tothe carrier’s attitude, vibration reduction and optimal multiple-objective. The mainresults achieved in this dissertation are summarized as follows.Dynamic model of the free-floating flexible space manipulator system isestablished, and the dynamic model is validated using ADAMS software. Then, theerror brought by modal truncation and the influence of free-floating carrier on theelastic deflection are also discussed. These works provide the foundation for the motionplanning of manipulator system.An improved motion planning approach of zero-disturbance to the carrier’s attitudein point-to-point manipulation is proposed. The zero-disturbance motion planningproblem is transformed into a non-linear parameter optimization problem. Combiningthe global search ability of genetic algorithm and the high convergence rate ofsequential quadratic programming, a pipelining approach containing the initial guess isproposed to solve the zero-disturbance motion planning problem. The globalconvergence ability and effectiveness of the pipelining approach are testified by solvingthe zero-disturbance motion planning problem of planar and spatial manipulator. Andthen, a motion planning approach based on translational zero-disturbance curves is putforward to improve the convergence ability of non-linear parameter optimizationapproach. Another approach for planning the speed along the zero-disturbancetrajectory is proposed, by which the optimal joint speed and manipulation time areobtained under the constraints of joint torque.A hybrid optimal approach for motion planning of vibration reducing inpoint-to-point manipulation is proposed. Considering the specialty of motion planningproblem in vibration reducing, a hybrid optimal approach combined the computingefficiency of Gauss Pseudospectral Method (GPM) and the accuracy of Direct ShootingMethod (DSM) is proposed. The effectivity of the hybrid approach is validated bysolving the motion planning problem of global vibration reduction and residualvibration reduction. Considering the uncertainty of initial state of flexible manipulator, amotion planning model involved the robustness performance index of residual vibrationis put forward, and the optimal motion for robustness performance index is obtained.A motion planning approach for trajectory tracking is studied. A motion planningapproach based on self-motion to minimize the attitude disturbance is presented. Thefeasibility of the proposed approach is validated in comparison with a planning method based on the pseudoinverse of generalized Jacobian matrix. Using the planningapproach based on self-motion and the hybrid approach employing GPM and DSM, themotion planning problem of vibration reduction in trajectory tracking manipulation canbe solved quickly and stably.An optimal multiple-objective motion planning approach of space manipulatorsystem is developed. The motion planning is proposed to take account of thedisturbance to the carrier’s attitude, the elastic vibration, manipulation time, and so on.Considering the specialty of optimal multiple-objective problem, the physicalprogramming method is used to convert the multiple-objective problem to asingle-objective one which expresses the designer’s preference, and is solved using thehybrid approach. The Pareto optimality of the solutions obtained by physicalprogramming is testified by NSGA-ΙΙ. And then, the designer preferred solutions tomultiple-objective motion planning problems of different manipulations are obtained.This dissertation proposes the motion planning approaches for minimumdisturbance to carrier’s attitude and vibration reduction, and also explores the optimalmultiple-objective motion planning model and approaches. Some developments intheories and methods have been attained. This study is closed to the application of spacemanipulators, and the presented motion planning approaches have a significant potentialapplication for the development of our space manipulator.更多还原