A study on tracking position control of pneumatic actuators using neural network

A tracking position control method for a linear positioning system composed of a pneumatic actuator and a 3-port proportional valve is proposed, and experimentally evaluated. The proposed controller has an inner pressure control loop and an outer position control loop. The position controller is based on a PID controller augmented with feedback linearization loop and neural network model, while a simple PID controller is used for pressure control. The influence of friction force and parameter change is regarded as a disturbance. The nonlinear relationship between the actuator velocity and acceleration, and the disturbance is coded on the neural network in training mode, and during operation proper input voltage that counterbalances the disturbance is calculated using the neural net model. Then, the outer loop PID controller is designed, assuming that the modeling error and the disturbance can be completely compensated, and, therefore, the actuator can be treated as a linear plant. Experimental verification indicates that the proposed controller significantly improves the tracking performance.