The Study Of Based-Neural Network Control In Rigid Robot Manipulator

One of the important manipulator operations is the control of the manipulator to track a given trajectory. Most commercial robot systems currently are equipped with conventional PID controllers due to their simplicity in structure and ease of design. Using PID control, however, it is difficult to achieve a desired tracking control performance since the dynamic equations of a mechanical manipulator are tightly coupled, highly nonlinear and uncertain. With the development of the modern control theories, more and more intelligent control ideas are applied to solving this problem. Up to now, intelligent control (IC) has been applied widely in the uncertain and nonlinear system. In this thesis, we aimed at the application of fuzzy neural network (FNN) technology in themotion control, and try to design a good control system for the rigidity robot manipulator.According to the structure character and the operation demand of trajectory of the Googol GRB400 industrial robot, the kinematics and dynamics analysis of robot manipulator was given. This robot is the SCARA articulated robot, which has four joints, and belongs to transport robot. Its joint 1 and 2 are rotated joints, and have two degrees of freedom (2-DOF) which complete the movement of the X-Y coordinate system. Joint 3 is straight joint, which finishes the kinesis of Z axle of coordinate system by lead screw. Joint 4 is used to adjusting the angle of tool, belonging to rotated joint. In this paper, the main job was the balanced problem of two main joints, and designed intelligent servo controller to make the dynamics quality better and to realize accurate trajectory.The article mainly concentrates on the following points: 1) A forward and inverse kinematics equation of four degrees of freedom (4-DOF) robot manipulator was set up by theknowledge of kinematics and dynamics of robot. Using the second dynamic mathematic modelling method—Lagrange functional balanced method, the dynamics model of 4-DOF robot manipulator was deduced.2) Then the basic principles of the robot application of fuzzy logic and neural network were introduced, in especial the structure and schematic of locally-attached neural network. The combinations of fuzzy logic and neural network were expounded, and the function equivalence of fuzzy logic and radial basis function (RBF) neural network was pointed out.3) In order to improve the tracking control performance under uncertainty, this paper presents a new intelligent hybrid control scheme for the manipulator, which is a RBF fuzzy-neural network multi-variable controller. The RBF fuzzy neural network multi-variable control algorithm has been brought forward, which separately controlled two loops of robot manipulator joint 1 and joint 2. Meanwhile it may decouple this system, and eliminate or reduce each coupling loop by the coordination network which wasmade up of single-layer neural network. At the same time, it has been adjusted on-line that the central value and width of the network membership function.4) The proposed control scheme has been implemented on a direct driven rigid robot manipulator of Googol. The outlined experimental results on tracking control demonstrate the effectiveness and the robustness of the new controller.