【作者】 张东；

【导师】 张秋豪；

【作者基本信息】 沈阳工业大学 ， 控制理论与控制工程， 2018， 硕士

【摘要】 机器人是集机械、电子、液压、控制、计算机信息、传感技术、仿生及人工智能等多个学科相互渗透、相互交叉的综合性学科。工业机器人技术研究日新月异,与人类生活密不可分,在机器人研究过程中,提出了更多的理论方法和仿真工具,使机器人的动态性能和数据精度都得到了很大改善。但是,由于在实际应用中存在各式各样的影响因素,导致了机器人研究更加的复杂性和困难性。对此,迫切需要建立一种新的机器人运动学和动力学模型,以方便提高机器人的可操作性和动态特性,从而提高机器人的精度和性能。本文以一种自动投料工业机器人在酿酒生产实践中为研究对象,阐述了一种以旋量理论为核心的全新机器人动力学参数辨识方法。首先,利用李群李代数和旋量理论对刚体空间运动旋量进行了详细的描述,引出运动旋量、刚体变换、指数积公式等概念,建立了基于旋量指数积形式的六自由度自动投料工业机器人运动学的模型和机器人几何参数标定。其次,基于旋量理论与凯恩方程,提出一种自动投料工业机器人的动力学建模方法;从旋量理论建立的指数积公式和螺旋运动两个方面分别阐述了自动投料工业机器人雅可比矩阵方程的建立过程,然后介绍了基于凯恩方程的动力学建模方法,在此基础上对各关节相应的主动力旋量、惯性力旋量、偏速度旋量、广义主动力及广义惯性力等参数做出了释义,并基于达朗伯原理完成凯恩动力学方程的推导,对影响动力学的因素也做了详细介绍。最后,基于参数矩阵递推方程建立了辨识方程,概述了以递推最小二乘法为核心的辨识算法,并对矩阵的满秩情况进行了详细说明,得到了辨识模型;该方法在很大程度上简化了辨识模型,降低了辨识难度,具有很好的针对性,从而获得准确的动力学惯性参数,优化了机器人动力学模型和改善了机器人动态特性,提高了机器人的辨识精度和动态性能。更多还原

【Abstract】 The robot is a cross-disciplinary integration,including mechanical,electronic,hydraulic pressure,computer,control,information,artificial intelligence,bionics and other high-tech field.With the rapid development of science and technology,some theoretical methods and simulation tools have been put forward in the process of robot research so that the dynamic performance and data precision have been greatly improved.Due to various factors in practical application,the complexity and difficulty of robotic research are caused.In this regard,it is urgent to establish new robotic kinematics and dynamics model to improve the maneuverability and dynamic characteristics of the robot in order to ameliorate the precision and performance of the robot.Taking an automatically feeding robot as investigated subject in the practice of wine making,a new dynamic parameter identification method based on the screw theory is studied in the paper.Firstly,the motion screw of rigid body is described by using the Lie group,Lie algebra and screw theory in great detail.Then,the concepts of the motion screw,the transformation of rigid body and the exponential product formula are introduced.Subsequently,the six DOF kinematic model of automatically feeding industrial robot and the calibration of the geometric parameters of the robot are also established.Secondly,based on the screw theory and Kane’s equation,a dynamic modeling method of automatically feeding industrial robot is proposed.The Jacobi matrix of automatically feeding robot based on screw theory is expounded on two aspects of the exponential product and spiral motion.Then,the dynamic modeling method based on the Kane’s equation is introduced.In this regard,the screw of active force,the screw of inertial force,the partial velocity twist,the generalized active force and the generalized inertial force are defined.Therefore,the Kane’s dynamic equation is built by the D’Alembert’s principle and the dynamic influence factors are also introduced explicitly.Finally,the equation of identification based on the recursive equation of the parameter matrix is completed,and the identification algorithm based on the recursive least square method is summarized.Then,the non-singular matrix is explained in detail.Hence,the identification model is obtained.The method simplifies the model in a large part and the difficulty of identification optimizes with a good pertinence.Accordingly,the dynamic inertia parametersof the robot are obtained,the dynamic model of the robot is optimized and the dynamic characteristics of the robot are improved.What is more,the accuracy and dynamic performance of the robot are improved.更多还原