Predictive Functioal Control Of Manipulator Bsded On SVM And PSO

Six-DOF manipulator is a typical industrial robot, and it has been widely applied in the welding, painting and assembly process. Because of the characteristic of strong coupling, nonlinearity and time-varying, the core of the research project is to realize high precision path tracking and control.On the basis of Lagrange method, the dynamic model is established. By analyzing the model characteristics, predictive functional control which independent of model precision is chosen. Predictive functional control (PFC) is analyzed from the selection principle of basic function, the establishment method of prediction model, the strategy of rolling optimal. Thus the control strategy are determined.In order to simplify the controller design, generalized inverse system for the manipulator is introduced. Considering the real manipulator system is unable to be mathematically modeled, the pseudo linear system composed of generalized inverse system based on support vector machine (SVM) and the original system, which is decoupled.An improved PFC method is presented. On one hand, the prediction model is formed by SVM. On the other hand, the control values are obtained by the roll optimizing of particle swarm optimization (PSO). Then the six decoupled PFC close-loop controllers are designed. The simulation experiments with PFC and PID control are carried out and the results are analyzed as well.The experiments on position and track of6-DOF manipulator are individually done with PFC, IMC and PID control method. The results with and without load experiments show that PFC method is flexible as well as IMC and PID, and the high efficiency and robust stability are obtained.