【作者】 胡松华；

【导师】 强文义；

【作者基本信息】 哈尔滨工业大学 ， 控制科学与工程， 2009， 博士

【摘要】 基于空间机器人技术来开展在轨服务研究是近年来航天领域研究的一个热点。本文以863航天重大项目为背景,围绕空间目标抓捕这一关键技术,有针对性的研究了空间机器人抓捕目标过程中的路径规划、动力学参数辨识问题,为了路径规划算法、辨识算法演示任务的验证,设计了空间机器人的动力学仿真系统和机械臂中央控制器的软件系统。研究内容主要包括以下几个方面:受地面重力因素影响,空间机器人地面运动实验很难准确反应其在微重力条件下的真实动力学特性,而数学建模可解决这一问题。为此本文结合空间机器人系统地面仿真验证需求,在考虑可移植性和计算效率的基础上提出了一套动力学建模仿真方案。该方案中,空间机器人的动力学方程基于拉格朗日方程和牛顿欧拉方程推导得到,针对动力学非线性项给出了一种数值计算方法,另外还分析了重力梯度对空间机器人动力学的影响并提出了减小的方法。基于该建模仿真方案设计了空间机器人数学仿真系统和半物理仿真系统中的动力学部分。空间机器人在轨服务通常可分为机械臂接近目标、手爪抓捕目标、抓捕目标后整个系统的稳定、对目标进行在轨维护操作四个阶段。目标的接近是关键的第一步,在接近目标过程中,机械臂运动会改变基座姿态,接近终止时刻机械臂的状态会影响抓捕时碰撞冲击对基座姿态的扰动,这会影响卫星的能源和通讯。为解决这一问题,提出了一种接近路径协调规划方法,规划机械臂关节轨迹和基座飞轮角动量使二者相互协调,确保接近过程基座姿态无扰动,同时可减小目标抓捕过程中发生接触碰撞后基座的姿态变化。抓捕目标后空间机器人负载发生变化,即等效的末端杆动力学特性发生了改变,这会影响抓捕后的机器人控制和规划,因此首先需要辨识其动力学参数。提出了一种基于惯量重分布的动力学参数辨识方法,即机械臂运动至不同臂形使已知参数部分获得不同的惯量分布,根据不同的动力学响应辨识出未知部分的参数。该方法不需要加速度级的测量,只需要用到位置级、速度级的测量结果,而且待辨识参数的结果采用最小二乘法求解,计算量小,因此具有很高的实用价值。文中还分析了测试误差对辨识精度的影响,为辨识算法的任务实现提供参考依据。目标抓捕后追踪星基座的姿态会产生一定偏差,机械臂关节和基座可能残留一定的(角)速度,为了进行在轨服务操作,需要将目标停靠在指定位置并调整基座的姿态。本文提出了一种基于微粒子群优化算法的非完整路径规划方法,并将该方法应用到抓捕后的目标停靠和基座姿态调整上。首先推导了系统动量不为零的自由漂浮空间机器人运动学,然后对机械臂关节函数进行了参数化,利用微粒子群优化算法求解得到机械臂的运动轨迹。该方法规划出的关节轨迹平滑、满足关节位置、速度、加速度的约束,同时还可限制机械臂运动过程中对基座的扰动角动量。为了充分演示验证空间机器人在轨服务的操作功能,本文有针对性的进行了在轨服务技术扩展实验任务设计,并完成了对在轨操作任务具备支撑功能的中央控制器软件系统设计。通过结构框架优化和任务动作分解,明确了界定了任务间耦合关系和工作模式转换关系,实现了对高复杂度在轨操作试验高效的数据管理和指令管理,在此基础上,重点设计了应用程序功能模块,给出了其功能的相关算法。针对本文提出了目标动力学参数辨识方法和非完整路径规划方法设计了扩展实验任务,并在该软件系统上进行了实现。更多还原

【Abstract】 In recent years,the research for on-orbit service using space robot system has been highly focused in aerospace field.Based on 863 aerospace research projects,this dissertation made a research in the problems of path planning and dynamic parameter identification in the course of target capturing.Dynamics simulator and software system of the central controller have been designed in order to demonstrate the tasks and algorithm for path planning and parameter identification,.The study includes the following aspects:Under the disturbance of the gravity,the ground experiment of space robot can hardly reflect the dynamic characteristic in micro-gravity environment.However,the embarrassment can be overcome by mathematics modeling.Integrating with the needs for the validation on the earth,a set of strong migrative and efficient scheme for modeling and simulation of space robot is proposed.According to the Lagrange equation and Newton-Euler equation,the dynamic equtions are developed and its nonliear term is derived with a special numerical method.Futhermore,the dynamic impact of gravity gradient on the robotic system is minimized after series of analysis.In order to validate target capturing capability,mathematics simulator and hard-in-loop simulator are established based on on-orbit mission scenario.On-orbit services using space robot can generally be divided into four stages: target approaching,target capturing,system stabilization and target maintenance.Approaching is the most important phase,the base attitude is disturbed because of the motion of manipulator and collision at the time of capturing,and it is very important for the energy and communication.A coordinated path planning algorithm is proposed to solve these problems.The trajectory of manipulator joint and the angular momentum of reaction wheel are planed,they are coordinated each other.After the capture operation is performed,the payload increment will induce a change on the equivalent dynamic parameters of the end body,naturally the dynamic parameter identification is required before space robot controller design.A kind of dynamic parameter identification method based on inertia re-distribution is proposed,the different manipulator configuration will generate different inertia distribution on known parts,then the dynamics parameters of unknown parts can be identified.As the remarkable advantage of this method,only velocity and position level measurement is necessary for the identification,more over,the calculation for identification results can be simplified by adopting least square method.In order to provide a reference for the parameter identification applications,the influence of measurement error on the identification accuracy is also analyzed.Since the angle velocity remaining on the post capture manipulator and platform,the base attitude should be regulated with captured target docked in a given location.Based on the Particle Swarm Optimization algorithm,a kind of nonholonomic path planning method for the target docking and platform attitude regulation is proposed.Firstly the non-zero momentum system kinematics of free-floating space robot is developed,and then the joint function is parameterized,by using the Particle Swarm Optimization algorithm,the manipulator trajectory can be obtained,the planning trajectory acquired by the above method is smooth enough,which satisfies the constraints about joint’s position,velocity and acceleration,furthermore during the manipulator movement,the disturbance of angular momentum on the base also can be assigned before the trajectory planning.In order to fully verify the on-orbit operation functions,the corresponding on-orbit missions are programmed for the space robot system,and the mission supporting software are also designed based on space level micro-processor.By configuration optimization and task decomposition,the coupling relations between tasks and the switching relations between task models are distinguished clearly,which enable the software system a highly efficiency data and command management function for the complex on orbit operations.Application function module is an emphases and the algorithms are given.Two extended missions are design base on the algorithms which are proposed in this thesis,and they are demonstated with the software.更多还原