Research On Feedforward Control Method Of Robot Based On Hybrid Inverse Dynamic Model

Industrial robot is a complex system with strong nonlinearity,large range of inertia variation and multi-domain coupling,so that it is quite difficult to realize high-speed and high-precision motion control of the robot.Feed-forward control based on robot inverse dynamic model is generally believed to overcome the nonlinear time-varying characteristics of robot to a certain extent,and effectively improve the dynamic control quality of robot at high speed.However,the existing research on this control method still has the following deficiencies: 1)The dynamic characteristics of the robot are complex,and the analysis and modeling of robot are lack of simple and efficient mathematical support,which leads to the low efficiency of modeling and calculation;2)Control-oriented inverse dynamics modeling of robots firstly requires analytical solutions of high-order derivations of kinematics of each joint,but it will introduce large deviations if using the conventional numerical difference calculation method;3)The imprecise inverse dynamics model has a significant impact on the control performance of robot,so that the traditional inverse dynamics feedforward control design and verification methods need to be improved.Therefore,combined with the above problems,this thesis mainly studies the serial robot dynamic modeling and feedforward control theory and technology based on Lie group and Lie algebra.The main research contents and results are as follows:The basic theory of Lie group and Lie algebra for robot dynamics analysis is studied,and the related inferences are proposed.The related theories and inferences are applied to the modeling of robot kinematics and dynamics,and the concise kinematics and dynamics equations are derived,which simplifies the derivation and calculation process,and a new parameter linearized dynamic equation is obtained.Based on the linearized dynamic equation,a parameter identification framework based on iterative learning is proposed.The feasibility of the parameter identification framework is verified by simulation results.The control-oriented inverse dynamics calculation method of rigid robot and the feedforward control method based on hybrid dynamic model are proposed.Based on the exponential product formula,the Jacobian matrix of robot is derived,and a recursive calculation method for the first derivative of Jacobian matrix is proposed,which can be applied to the analytic calculation of the first and second derivatives of joint displacement,and provides effective support for the accurate calculation of inverse dynamics of rigid robot.In order to reduce the influence of parameter uncertainty and nonparametric uncertainty of inverse dynamics model on control performance,a feedforward control method based on analytical and data-driven hybrid dynamics model is proposed.Through simulation and experimental analysis,the correctness of the robot inverse dynamics calculation method is verified,and the analysis results show that the proposed control method can significantly improve the trajectory tracking accuracy of the robot.Based on the inverse dynamic model calculation and feedforward control of rigid robot,the robot inverse dynamics calculation method considering joint flexibility and feedforward control method based on hybrid dynamics model are further studied.A recursive method to calculate the higher-order derivation of Jacobian matrix is proposed,and it is applied to the calculation of the high-order derivatives of robot joint kinematics.By applying Lie group adjoint mapping inference,a recursive inverse dynamic model of flexible joint robot is obtained,which is more efficient and easier to program.In order to reduce the influence of the inaccuracy of the dynamic mechanism model on the control performance,a feedforward control method of flexible joint robot based on analytical and data-driven hybrid dynamic model is proposed.The correctness of the inverse dynamics calculation method and the superiority of the feedforward control method based on the hybrid dynamic model are verified by numerical simulation.A more perfect and effective multi-domain integrated system model of robot is established,which provides a reliable digital analysis method for virtual verification of robot control algorithm.Considering the multi-domain coupling characteristics of industrial robot,the hierarchical progressive modeling method of robot multi-domain system is studied,and the multi domain simulation analysis platform of robot is constructed.For a certain type of robot,the multi-domain system model is established,and different control algorithms are simulated and compared based on the model under different working conditions.The simulation results show that the feedforward control method based on the hybrid dynamic model proposed in this thesis can significantly improve the trajectory tracking accuracy of the robot,reduce the influence on the control performance caused by the inaccuracy of the dynamic mechanism model.Finally,the experimental study is carried out,and through the comparative analysis of the theoretical results and the experimental data,the correctness of the robot inverse dynamics calculation method and the robot multi domain system synthesis model is verified.