Self-adaptive Trajectory Control of RBF Neural Network SCARA Manipulator

The SCARA manipulator is precision electronics, aerospace, automotive and otherindustries commonly used assembly and handling equipment. During the movementvelocity, acceleration, and jerk and other factors will affect the smoothness of operationof the track so that the track quality and accuracy of the assembly is greatly reduced.This article will introduce intelligent algorithms RBF SCARA robot control system.The main contents are as following:(1) The structural characteristics of SCARA manipulator are studied, kinematicmodel is analyzed. The forward kinematics is realized through the matrix transformation,and the inverse kinematics are deduced by closed geometric method. Then dynamicsequations is completed for the Lagrange equation. On this basis, dynamic error and thestatic error of the system itself is analyzed.(2) According to the different requirements of the performance of the controlsystem, the traditional control and intelligent control strategy in the field of robottrajectory tracking control of the control precision and applicability is discussed. RBFneural network control algorithm is proposed. It is used to eliminate the complexSCARA manipulator which is a complex nonlinear control object, uncertainty dynamicerror compensation and the approximation capability of the network itself. And thedesign of the neural network controller has the ability to learn by using digitalsimulation technology. Simulation results show that the proposed control strategy to theunknown external disturbances SCARA manipulator control system is effective, andmore accurate trajectory tracking, and in network training times and error convergencespeed very quickly to achieve satisfactory tracking performance.(3) In view of the initial value selection of imprecise network, it may cause that thecharacteristics of the track effect is not ideal. A parallel feedforward neural networkcontroller and the linear proportional feedback controller combines differential modewhich is established. Then, it applies to the above adaptive trajectory tracking controlmodel. Network structure and feedback control feedforward signal common drive eachjoint.(4) RBF neural network control algorithm is introduced Matlab/Simmechnicsphysical modeling to verifiy correctness and feasibility of SCARA robot dynamicsmodel and dynamic disturbance compensation control algorithm for RBF. Thesimulation results can not only fast and stable system through a given path constraintpoints, and to ensure that the movement of the joints continuous smooth curve.