| When robot moves, time-varying and nonlinear characteristic of the physical parameters of robot joint will be shown, so the dynamic model of the robot that is built at fix state can not describe dynamic characteristics of the robot under moving state. The research about physical parameters identification of moving robot joint begins to be paid attention. In this paper, several time-varying physical parameters identification methods of robot joint are presented based on wave transmission method, time sequence method, fuzzy method. Dynamic modeling of flexible joint robot is researched, and the computing method about force and response of robot joint is presented.The dynamic model of robot is the base of identifying physical parameters of robot joint. In the paper, the dynamic characteristics of robot with nonlinear joint is discussed. Dynamic equations of robot having rigid beam and flexible joint are deduced and dynamic equations of robot having flexible beam and flexible joint also are done. Through a simulation example, the effect to error of robot end produced by nonlinear characteristics of robot joint is discussed. The content in second chapter is the base of researching physical parameters identification of robot joint in next chapters.During identifying physical parameters of moving robot joint, the response of robot joint must be used. In practice project, generaly speaking, it is difficult to test response and force of robot joint, so they must be computed by other method. In the third chapter, the two methods about computing response and force of robot joint are presented. One is that response of robot joint is computed by testing point and math equations of testing point and joint point, the math equations of testing point and joint point are gotten by unit pulse response between testing point and joint point. Another method is deduced based on substructure. A simulation example to two solid beams is used to testify the method.In the paper, three online identification methods about time-varying physical parameters of robot joint are presented.(1) The time-varying parameters identification method of robot joint is presented based on wave transmission theory. The characteristics of vibration wave are gotten through wave decomposing theory. The wave mode of each unit is built. The mapping relation between dynamic stiffness of joint and amplitude of wave mode is built based on the relation between displacement and force of robot joint and amplitude of wave mode. During identifying physical parameters of robot joint, the mode parameters of robot are used to compute wave amplitudes, so subspace decomposing method about identifying mode parameters of robot is presented. The time-varying physical parameters identification method of robot joint is verified by simulation example.(2) Nonlinear dynamic model of robot joint is built by dynamic model of whole robot. The nonlinear dynamic model of robot joint is dispersed to nonlinear time sequence mode based on linear time sequence ARMA model and the conversion between differential equation and nonlinear time sequence model is discussed. At last the coefficient of the nonlinear time sequence equation is gotten. The feature matrix of the equation is computed by Singular Value Decomposition method, so computing speed of the method can be advanced, and computing time can be cost. The simulation to a mass-spring system is carrieded out to verify the method.(3) When robot moves, the dynamic model of robot joint is complex,uncertain and nonlinear. In the paper, the dynamic model of robot joint is expressed with fuzzy model. The online self-adaptive adjusting method of fuzzy clustering is presented and computing method about online interpolation fuzzy membership is researched. The physical parameter of robot joint is computed real-time by using vibration signal of robot. In the chapter, a cantilever beam system is used to verify the method.At last, the time-varying physical parameters identification methods of robot joint presented in the paper are validated by experiment. In the experiment, a 3-DOFs robot is taken as research object. The physical parameters of the move robot joints are identified real-time by using wave transmission method and fuzzy identification method respectively. The values and changing trend of physical parameters of the robot joint are gotten. The identification results are discussed. The experiment results show that the methods presented in the paper have steady computing result, high dependability and applied value in the practice project.
|
PDF Full Text Request