Robotic Manipulator PD Control Driven By Virtual Unmodeled Dynamic

Industrial robotic manipulator is a typical representative of electromechanical integration equipment, which is more widely used in today automatic processing lines. Its technologies are relative mature and its scientific research value is promising. High precision servo control of manipulators has always been an important research direction of mechanical and electrical control system. But at present, industrial field often use linear feedback PD control method on manipulator servo control. However, the accuracy of individual axis positioning can’t satisfy the requirement of high precision and its control effect is not satisfactory. While nonlinear control methods often need complex design process, large amount of calculation or highly rely on model accuracy. In order to solve this problem, aim at manipulator system, researching on a kind of high accuracy and nonlinear servo control method, which has less amount of calculation and easier to be realized, has important research significance and application value.To solve the above problems, this article apply the PD control method with virtual unmodeled dynamic compensation to manipulator servo control based on data driven, relying on the state key laboratory of integrated automation of process industry. The method has obtained well control effects, the main research contents are as follows:(1) Study industrial manipulator, the complex nonlinear servo control mechanical and electrical object, analyzes servo control methods on manipulator, and points out the difficult problems of high precision servo control. Meanwhile, established the hybrid dynamic model of manipulator for angular displacement servo control, which include:manipulator body model and the model of drive system, and then get the discrete model for controller design.(2) For industrial manipulator dynamics systems with strong coupling between each joint, nonlinear, strong friction disturbance etc. comprehensive characteristics, using industrial linear PD feedback is hard to get ideal control effects. In addition, neural networks and other online learning methods with complex calculation and large networks are far away from practical application. So this article puts forward a kind of PD control algorithm driven by virtual unmodeled dynamic, which consists of the PD controller, linear model, virtual unmodeled dynamic and friction compensator. This paper also analyzes the stability and convergence of the control method in theory.(3) Further complete the manipulator fast prototype control system experimental platform. Make experimental comparisons for the proposed control method and conventional linear PD control algorithm, PD with pole placement algorithm, PD based on disturbance observer compensation algorithm, and nonlinear friction adaptive compensation algorithm. Then puts forward some quantitative evaluation index for servo control performance. The experimental results show that the proposed control method can effectively improve the performance of manipulator servo system, reduce tracking error, and improve steady-state accuracy.