Design And Research Of Intelligent Controller For Tracking Robot Trajectory

This paper discusses the trajectory tracking robot research of uncertainty robotmanipulators, analyzes the uncertainties that exist in the robot control. Firstly, thetypical of industrial control of PD control algorithm is discussed, and the PD controlalgorithm design process is analyzed, the simulation results are given. Consideringthe simplified design process of the PD control method, this paper ignores theuncertainties, putting forward the feedforward robust control PD control algorithm.Discussed the characteristics of feedforword control and the detailed process and thesteps of controller design, the stability of the system is proved. And for the nonlinearand strong coupling of robot control system, the robust control scheme is puttedforword. Introduced the related knowledge of the robust control, in view of the upperbound of the uncertainties, the robust controller is designed.As we know, the neural network has nonlinear mapping capability, so it hasbeen widely used in nonlinear control. Robust controller which is designed based onneural network is put forward in this paper. Using the neural network to estimatesexternal interference of the neural network approximation for produced by the erroras an external disturbance, the designed robust control rules to eliminate the error.The simulation results prove the neural network robust controller is effective of andhas good dynamic performance and robustness. Further research, the neuralnetwork-based adaptive sliding mode control scheme is put forward. For existing ofuncertainties, using neural network approximation respectively, using sliding modevariable structure controller and adaptive controller to eliminate the producedapproximation error. This paper gives the specific design steps, the property withdirectly for the use of neural network to estimates of the uncertainties of the algorithmare compared in the simulation results, the neural network adaptive sliding modecontrol scheme has its advantages and reliability in the tracking error, and thedynamic performance, control the stability of the moment.This paper finally, considering the robot system with model uncertainties andnonlinearities, an adaptive controller based on RBF neural network and backsteppingis proposed.No longer simply rely on neural network approximation ability, but using backstepping method to design the system step by step. Then the quantity controlusing neural network to estimates of the uncertainties, making full use of systemknown information design is derived. The simulation results compares with theproperty of the scheme using the neural network alone, proves the effectiveness ofthe control scheme based on neural network and backstepping is correct.Finally, the paper summaries the research work and proposes further research.