Research On Adaptive Terminal Sliding Mode Of Manipulator Based On Neural Network

Manipulator is constituted by mechanical body, controller, servo driver system and detection sensor device, which has characteristics of human-simulated operation, automatic control and reprogramability. It is a electromechanical equipment that can complete various operations in three dimensions.The use of manipulator has become increasingly prevalent and far reaching. Manipulator almost can do whatever human can do. Manipulators are playing an increasingly important role in daily life, industrial and agricultural production, oceanographic and space exploration. Manipulator is a very complex multi-input multi-output nonlinear system. It has dynamic characteristics such as time variation, coupling, nonlinear. The controlling problem of manipulator is to make the each joint and the end effector can track the given trajectory with ideal dynamic quality and stable at the specified location. In practice, many factors can cause interference and uncertainty to the manipulator system. This uncertainty and disturbance may cause system instability. Traditional robust controller has a certain vulnerability which may result in decreased performance and even the instability of the closed-loop system. Sliding mode variable structure control technology was founded as soon as the cybernetics was raised. Because of its strong robustness, it was given close attention. Sliding mode control technology is widely used in the manipulator control. Therefore, studies with terminal network adaptive sliding mode controller have important theoretical and practical significance.In this paper, we will study in the manipulator neural network adaptive terminal sliding mode control. The main work is summarized as follows:First of all, I studied manipulator neural network terminal sliding mode control problem. Using the neural network identified upper bounds of uncertainties. And by designing the sliding surface with saturation function, terminal sliding mode controller was designed. Because of the design process utilizing Lyapunov function to ensure the stability of manipulator systems, we propose a feasible adaptive neural network terminal sliding mode control algorithm. Then for the model uncertainty of the nominal model, we identified the neural network and designed a terminal neural network sliding mode controller which can identify the upper bound of the interference and uncertainty. And simulation results show the effectiveness of the method.Secondly, I studied the inversion technique to design the terminal sliding mode controller. Normal linear sliding surface can not converge to zero in a finite time. Therefore, the introduction of nonlinear terms in the sliding surface and inversion technique constitutes a fast terminal sliding mode controller. Combined with neural network to identify the uncertainty bound manipulator system, I designed a new fast terminal sliding mode controller. And simulation results show the effectiveness of the method.Finally, the main results are summarized. In addition, we discuss some issues worthy of further study in the future about terminal sliding mode control of the manipulator. And there are some inadequate methods mentioned in this article.