| With the development of science and the need of human, robot has already developed to distant space. Because space environment is complex and danger, for example space microgravity, high vacuum, strong radiation and the existence of planetary body, so it is difficult for human to accomplish task, space robot appear in spacewith its unique advantages, replace human do difficult task, which are highly valued by all world scientists.In actual work, there is uncertainty in the system of space manipulator. The control effect is worse and the system is not stable used control strategys based on system dynamics such as coordination control and calculation torque control. In order to solve the problem of trajectory tracking of uncertainty and attitude-controled manipulator in joint space, two kinds of sliding mode control methods are designed in this paper:First, based on Terminal sliding mode control method, the uncertainty of system and the interference of the upper bound are learned by the neural network. The chattering problem in system is reduced because symbols function is replaced by saturated function. Therefore, the tracking error converges to zero in a limited time. For further improving the method, two RBF neural network is respectively used to approach the nonlinear function of system and the uncertainty bound. With the improved method, it is not requirement to compute complex dynamic model. The comprehensive control of pedestal stance and mechanical manipulator joints is achieved in the system which exists uncertainties and external interference.Second, the edge layer thickness of saturated function is unable to accurately determine in actual with the first method. In order to solve this problem, a kind of inversion terminal sliding mode control method is raised with combining of the in-version method and sliding mode control method. The influence of system produced by uncertainty is effectively eliminated and the stability tracking of system trajectory is implemented. |
PDF Full Text Request