Reach Based On Rbf Neural Network Of Pneumatic Position Servo System

Pneumatic has the advantages of Simple structure, no pollution, reliable work, transmission speed and free maintenance, and it has been widely used in industrial automation and robotic drive system. However, due to some of the inherent characteristics of pneumatic technology, such as great compression of gas, strong friction force of air cylinder and flow of non-linear when gas through the valve port, it difficult to achieve the desired results of pneumatic control accuracy and stability, and it limited the wider application of pneumatic technology. Therefore, reach to pneumatic position servo system, take appropriate control methods to achieve precise position control is one of the important projects in pneumatic field.Based on the mathematical model of pneumatic position servo system which composed of pneumatic servo valve and rodless cylinder, and through the study of neural network theory and pneumatic position servo system simulation studies based on RBF neural network, it realized precise control of rodless cylinder position.This paper summarized and explained the development and current situation of pneumatics and servo control at home and abroad, and presented situation of the current issues and research content of this article.The three-DOF manipulator based on pneumatic pneumatic position servo system's hardware and software composition and working principle was introduced, non-linear mathematical model of the system was established and the characteristics of the mathematical model was described.On the study of control strategy of the pneumatic position servo system, it proposed control strategy used in this article:RBF neural network-PID compound control. Through applications of RBF neural network, it realized the dynamic adjustment of the PID parameters and makes the PID parameters in the best condition when the system was running, and it could ensure stability of the system.MATLAB/Simulink and AMEsim software was used in this paper to make simulation research of the system. In Simulink environment, mathematical model of the system was established and simulations was made by using RBF neural network-PID controller. Simulation results showed that the RBF neural network-PID compound control significantly improved the characteristics of the system campared with the traditional PID control method, bandwidth up to 5 Hz, tracking accuracy of 95%. In addition, the physical model of pneumatic position servo system was built by AEMsim software, and simulation was made in simulink environment. The simulation results also showed that control precision and robustness of the system has been improved and joint simulation of AMEsim and Simulink was realised.