Research On Sliding Mode Control For Multi-joint Robot

Sliding mode variable structure control has stronger robustness, especially suitable forsolving the problem of trajectory tracking of robot manipulators. However, its application anddevelopment are limited by the chattering phenomenon of the SMVSC systems. In order toalleviate chattering, merging SMVSC with other intelligent control methods are efficientapproaches, such as fuzzy control, neural control, backstepping control and so on. The mainworks of this paper are as follows:First of all, on the basis of sliding mode control combined fuzzy control andbackstepping control. Respectively are fuzzy sliding mode control based on the backsteppingcontrol and the fuzzy adaptive sliding mode control gain adjustment based on thebackstepping control. The first method is based on the backstepping of the sliding modecontrol with fuzzy compensate the friction term of model, thereby eliminate chatteringeffectively. The second method is using adaptive fuzzy systems to approximate the gain ofcontrol law and eliminate the chattering of sliding mode control. Simulation results show theeffectiveness of the proposed two methods.Secondly, study the approximate characteristics and basis of RBF neural network. Fortracking a desired trajectory problem of multi-link robots with modeling error and externaldisturbance, proposed two control methods. Respectively are multi-joint robot RBF networksliding mode control based on the model of the approaching and sliding mode control robotbased on the radial basis network minimum parameter learning method. The first method isusing RBF network approach the model information of every join. The second method is touse a single parameter. It does not need model information. It can replace the weights of theneural networks and adaptive control based on a single parameter. Simulation results show theeffectiveness of the proposed two methods.Finally, In view of the chattering of the sliding mode controller, improving the globalsliding surface, and using the robot’s physical characteristics, adding a low pass filter at theoutput of controller to eliminate chattering phenomenon. Simulation results demonstrate theeffctiveness and fasibility of the proposed control strategy.