Dynamic Performance Oriented Industrial Robot Control Technology Research

With the increasing demand of industrial robots in the automated production areas such as mechanical processing, automotive manufacturing, assembling, palletizing and others, robots’ working efficiency response speed, tracking accuracy and settling time and other performances have been hot issues in recent research. In order to improve the dynamic performance of the robot, the palletizing robot dynamics was modeled. And based on this, the related issues of robot dynamic performances oriented driving system designing, robot controller design based on linearized dynamics model and robot vibration suppression were reaserch. Moreover, the control structure was designed based on industrial PC and high speed communication bus and the corresponding performance test and experiment were implemented in our palletizing robot control system.The kinematic and rigid dynamics models for handling robot were built. Considered joints and links elasticity, the rigid-flexible coupling model of robot was built based on Lagrange equations and Lumped Parameter Methods. The kinematics was regarded as constraints according to the characteristics of the robot dynamics equations. The original non-linear differential equations were extended to differentialalgebraic equations with algebraic constraints, so that the rationality and accuracy of solving the inverse dynamics were both improved. The state variables for solving dynamics equations satisfy dynamics and kinematic constraints. The differential index of rigid-flexible coupling dynamics was reduced by differentiating algebraic equations constraints and the introduction of dummy variables. Therefore, the problem was converted to the initial value problem of nonlinear differential equations.For designing robot driving system, the impact of driving components for the overall system performance was analyzed. According to the performance requirements, the optimization model was built with the objective function of robot working efficiency and natural vibration frequency. Furthermore, the constraint conditions included peak and rated torque of the motor, life expectancy of reducer and load weight ratio, etc.. To solve the discrete variable optimization problem, the mapping relationship between discrete optimal design variables and system properties were established. The performance indexes and constraint conditions were simulated by dynamics modelling. Furthermore, the discrete variables optimizing issue was solved using mixed variables genetic algorithm. The optimization results show that optimization model and optimization methods are reasonable applicable, and the corresponding dynamic performances have been improved effectively.According to the robot vibration patterns, the vibration of robot was divided into two forms: Residual vibration and vibration in the movement. The robot residual vibration was analyzed first, which motion was depicted as free vibration equations in the free vibration state. The time domain response of vibration equations was obtained based on the modal analysis theory. The main impact factors of residual vibration were attributed to the trajectory position and velocity errors at the stop time of the robot movement. From the perspective of control system, the impact factors of vibration during the robot movement caused by the reference trajectory were summarized as the trajectory’s continuity at the start and endpoint as well as frequency characteristics. The performance index function was built based on the flexible joint dynamic equations, and the residual vibration suppression motion planning problem was regarded as functional extremum problem. The functional extremum problem was converted to solve the boundary value problem of ordinary differential equations based on the Pontryagin maximum principle. As to the reduction of vibration during the movement of robot, the optimal trajectory results were fitted by Fourier series under conditions that the velocity, acceleration boundaries are continuous at the start and endpoint. In order to reduce the probability of exciting the natural frequency of robot, the high frequency harmonics included in feedforward torque was decreased as much as possible under the premise of fitting accuracy guaranteed. Results show that the vibration suppression trajectory planning strategy can effectively reduce the amplitude, time of residual vibration and the deformation of flexible joint during movement.For the model-based robot control, first, the nonlinear characteristics of dynamics of flexible joint robot were analyzed in the whole working space. And then, the robot convex polytopic model was established using the high order singular value decomposition method. The effectiveness of the linearization model was verified through calculation and simulation, which show that the convex polytopic model obtained by this method can accurately describe the nonlinear dynamics characteristics of flexible joint robot in whole workspace. Based on feedback linearization controller design method and optimal control theory, the optimal control problem was transformed into the problem of solving LMIs by solving the common positive definite solution of the group of algebraic Riccati equations which corresponds to each vertex in the linear time invariant system of robot convex polytopic model, and the optimal gain-scheduling controller was obtained by solving the corresponding Linear time invariant system.For the implementation of the control system of the palletizing robot, the motion control task was decomposed first. And then the fundamental motion control software was developed according to the state machine. Accordingly, the control system was proposed based on high-performance PPC and high-speed communication. The repeatability precision and dynamic performance of the experimental prototype were tested according to the national industrial robot test standard. Finally, the validity of the proposed partial control algorithm was verified in the engineering application.