| Industrial robotic manipulators are widely used in industrial manufacturing and assembling lines, which contributes greatly to the increase of productivity. At the same time, industrial robotic manipulators can be used to deal with many different jobs. They can fulfill many kinds of manufacturing, assembling and moving jobs with low costs. Although the research of industrial robotic manipulators and their control algorithms has had a long history, it's difficult to apply many algorithms to real plants because of the high non-linearity, high coupling, time variation of the system, problem in model accuracy, computation consumption and equipment costs. In addition, there are many disadvantages, which hinder the robotic manipulators from doing jobs requiring higher accuracy and speed, in the algorithm that used in the control of commercial robotic manipulators.Iterative learning control algorithm is a good approach to control industrial robotic manipulators because it can let the robotic manipulators follow the expected trajectory by iterative learning without the accurate model of the system. In real robotic manipulators there are many constraints such as max torque of motor, rotation angle of joint, difference in length of the working time span. These constraints are not in accordance with the assumption of iterative learning algorithm and so bring difficulties to the application of it. Thus, it is a good subject to investigate the iterative learning algorithms' application in robotic manipulators. So that it can be convergent in those conditions under constrains and can control robotic manipulators properly to realize trajectory tracking well. In this paper, several iterative learning algorithms were presented with astringency testifying and system simulation to indicate the astringency and practical effectiveness of them.The following are the contributions of this paper:Robotic manipulator is a time-varied system. So it's difficult to achieve good convergence effect and highly convergent speed if the constant iterative learning operators are used. As a consequence, one proper time-varied operator should be adopted instead. In this paper, modifications were made to the assumptions of Avrachenkov's[31] quasi-Newton iterative learning rule of robotic manipulators. At the same time, astringency testifying was presented. Algorithm A was obtained after modifying quasi-Newton iterative learning rule with feedback. With references to Tae-yong Kue Kwanghee Nam and Jin S.Lee's[32] literatures, algorithm B was obtained after modifying algorithm A.The output signal of industrial robotic manipulator controller is correspondence with the torque provided by each joint. In practice, outputs of motors are constrained by max torque, so do outputs of controller. As a result, controllers cannot output signals according to the computation result of algorithm, original iterative rules are affected, and it even affects the convergence of the algorithm. In this paper, discussion of this problem was presented. Based on continuous time system, convergence discussion and testifying were made to iterative learning control algorithm under the condition of constraints. Then algorithm A and algorithm B that mentioned before are testified that they can be used under the conditions of that controller output has constraints.Each joint of industrial robotic manipulators cannot rotate to any angle because of mechanical constraints. These constraints are not indicated in dynamics functions of robotic manipulators. Constraints of objects' output change the quality of control object of iterative learning control algorithm greatly. So it's essential to discuss the convergence of the algorithm again. Discussion and testifying were made to the convergence of the algorithm under the condition of having constrains in objects' outputs. One new algorithm, which can maintain the convergent speed under such constraint conditions, was presented.Using iterative learning control in real job of industrial robotic manipulators, we h...
|
PDF Full Text Request