Control Strategies Study On Uncertain Robot Under Constrained Input

Robotic manipulators are a typical, time-varying, strong-coupled nonlinearsystem. With the further study of robotics technology and the development ofapplied area for robots, conventionally used control algorithms, such as PID control,have not met the requirement of these areas. The appearance of a good manyintelligent control schemes solved the nonlinearity in the robot system muchsuccessfully. Because of the conservative nature in designing the intelligentcontrollers, they perform very high amplitude and high velocity. However, they arehard to realize in practice, which confines the practical application of moderncontrol strategies in large extent.According to the above situation, a novel strategy, which considers theconstrained input, is presented in this dissertation. With the predefined rapidity ofthe close-looped system and high degree of accuracy, this subject can provide therobot control strategy with lower amplitude of the controller. The main study workcan be generalized as follows:On the basis of existed robust control strategy, the study situation of thecontrol strategies under constrained input is analyzed firstly. Secondly, from theconservative viewpoint, robot controller design is divided into three types, and thesignificance of studying control strategy under constrained input and several basicproblems have been proposed. Then, a type of filtering function satisfying therequirement of controller considering constrained input can be obtained, and takingthe typical hyperbolic tangential filtering function as an example, the controllerdesign and simulation in robots have been performed;the effectiveness of thiscontrol strategy has been demonstrated.In the design of constrained input controller, the normal robust controller basedon counting-force structure is decentralized in this dissertation. And with the baseof explicit form for force input, the nature of high amplitude output for normalrobust controller has been analyzed. Then, with total state feedback, roboticcontroller has been designed based on hyperbolic tangential filtering function, andthrough the simulation and quantitative analysis, the good property of the controllerhas been demonstrated. According the same idea, a constrained input controllerbased on gravity compensation and time-varying parameters has been designed, andthe control effect is perfect.Several existing modern intelligent control strategies have been analyzed in thedissertation. Against their common high control input requirement, the constrainedinput idea is put into the intelligent control, and fuzzy self-adjusting PD constrainedinput controller and constrained input fuzzy variable structure controller based onsystem status have been designed. Through simulation, it has been proved that thecontroller enables the robots to satisfy the control requirement of rapid speed andhigh accuracy under a smaller control input.Finally, against the factors affecting the practical control effect, such as thestrong-coupling the nonlinearity of 6-DOF parallel robot, the introduction of thedecentralized variable structure control strategy can realize the model decouplingand the compensation of unfavorable factors. And combined with the constrainedinput control strategy, input limitary decentralized variable structure controller hasbeen designed and obtained perfect dynamic performance, which provides thecomplex variable structure controller with better practicability and widerapplication domain.In order to demonstrate the constrained input control strategy in practicalsystem, the verified experiment has been done on the experimental equipment.Compared with the conventional control strategy, at the same time of increasing thetracking accuracy, the control input has reduced by 40 percent, and the controleffect is perfect.