Research On RBF Adaptive Sliding Mode Control Of Reconfigurable Manipulator

Reconfigurable manipulator can be rapidly replaced and assembled into different robotic configurations under conditions of certain circumstances,possessing the salient advantages of modular,high-precision and configuration adaptability.In practical applications,many uncertainties existing in system and the external disturbance,which will affect the performance of trajectory tracking.Therefore,the research topic of dynamic control method adapted to configuration changes of reconfigurable manipulator and effectively compensate uncertainties is imminent.Several new trajectory tracking control strategies are presented for solving the configuration-adaptive and compensation problems of the reconfigurable manipulators in joint-space control and task-space control,at the same time,theoretical deductions and simulations are also carried out.For solving the uncertainties problems of the reconfigurable manipulator and the trajectory tracking problems of the robot configuration changes in the joint-space control,a new robust configuration-adaptive control strategy based on RBF neural network is presented.RBF neural network compensating controller is designed to approximate the uncertain items and the unknown system input caused by reconstruction.In order to compensate the system approximation error and improve the anti-interference and the robustness of the changes of configuration,the robust term is appended to the above controller.For improving the performance of configuration-adaptive,an configuration-adaptive parameter adjustment law plus a robust term is also proposed based on the Lyapunov theory,and the system stability is verified subsequently.The substantial experiential results based on simulations show that the proposed robust configuration-adaptive control strategy based on RBF neural network can quickly achieve each joint module track desired trajectory accurately under the circumstances of the configuration changes and the system uncertainties.Considering the sliding mode control algorithm is better robust to parameter uncertainties and external disturbances of controlled object and adapt to the complexity of reconfigurable manipulators’ configuration changes,a control algorithm based on the fuzzy adaptive sliding mode gain control is presented.A sliding mode controller consists of sliding mode surface based on saturation function and approaching law based on exponential function is proposed for compensating the system uncertainties and the external disturbances.Considering the limitations of reducing the chattering of the sliding mode gain control and can not effectively compensate the time-varying uncertainties of system,a new sliding mode gain is designed based on the constant sliding mode switch term and the feedback term,and an fuzzy adaptive controller is real-time applied to adjust that.Subsequently,an configuration-adaptive parameter adjustment law is proposed based on the Lyapunov theory,and the stability of the closed-loop system is verified.Contrasting with robustconfiguration-adaptive control strategy based on RBF neural network,the substantial experiential results show that the control algorithm is more effective,not only quickly adapt to the robot configuration changes without adjusting any control parameters,but greatly reduce the chattering of the sliding mode control.In order to the select the configuration adaptive parameters conveniently,based on the analysis of the control performance with the configuration adaptive control parameter P in the case of invariant controlling parameters,the approximate optimization interval of P is derived.For solving the end trajectory tracking problems of the reconfigurable manipulators with uncertainties in task-space,a new adaptive sliding mode control strategy based on Radial Basis Function neural network(RBFNN)is presented.The dynamic equations of the reconfigurable manipulators is transformed from joint-space to task-space.The designed stable adaptive sliding mode controller based on RBFNN is used to approximate and compensate the unknown closed-loop system dynamics of the reconfigurable manipulators directly in task-space.The weight-adaptive adjustment law appended to the above controller is greatly to reduce the influence by the uncertain external disturbance as well as the approximate error of the neural network system.Based on the Lyapunov theory,the system stability is verified subsequently.The substantial experiential results that the tracking trajectory of end effector under the proposed method has preferable tracking performance,similarly.