Sliding Mode Variable Structure Control For Underactuated Horizontal Manipulators

Underactuated manipulator system is a typical second-order nonholonomic dynamic systems and isn't an entirely controllable system. The manipulator's motion control and planning is much more difficult than fully actuated robot arms and the smooth feedback control methods cannot be directly applied because the existence of the underactuated joint. Moreover, it is difficult to find a common and method with better control results. Therefore, the underactuated manipulator system control problems have attracted the worldwide attention of the academia and widespread concern.Hierarchical sliding mode control theory, which was emerged in couple years ago, is a sliding mode control method which use hierarchical strategy to design the actuated and underactuated joint's sliding surfaces respectively and combine it with nonholonomic constraints manipulator systems to achieve the effective manipulator control.This thesis proposes several new control system design techniques for stabilizing position control of the underactuated system. The dynamic model of 2-DOF underactuated manipulators is directly used in all researches. The main topics are as follows:Firstly, the paper presents a comprehensive, detailed, and complete analysis for the nonholonomic constraints system's research and development process. Then, the thesis gives a hierarchical sliding mode variable structure control method for the 2-DOF Horizontal underactuated manipulator and theoretically proves manipulator's Lyapunov stability and achieves its position control.Secondly, the paper presents an improved genetic algorithm, which based on time-varying crossover and mutation operator to achieve the controller's parameters design. The method improves the speed of algorithm convergence and bringing further controller parameters optimizations. Thirdly, a general design method of RBFNN hierarchical sliding mode control is described and a RBFNN sliding mode control algorithm is proposed for underactuated systems in order to solve the chattering of sliding model control and stability error problem. Then, The idea of hybrid coding method is proposed to enhance RBFNN genetic optimization of efficiency, and the use of adaptive genetic algorithm to achieve the RBF neural network structure and parameter optimization, solve the sliding control of the "chattering" and the steady-state controller error.Finally, a novel adaptive control algorithm which combines backstepping design and hierarchical sliding mode control strategy is proposed for underactuated systems to enhance anti-jamming capabilities. Moreover, the paper gives a detailed proof of the controller's Lyapunov asymptotic stability.Simulation results show that the control methods are all effective.