Improved Fuzzy Neural Network Control for a Pneumatic System Based on Extended Kalman Filter

Although pneumatic systems are used in many applications such as robotics and manufacturing field, accurate control for such systems is difficult to be achieved due to their inherent nonlinear dynamics. This paper presents the favored results of fuzzy neural network (FNN) control for a pneumatic system based on extended Kalman filer (EKF). To optimally design a FNN controller for the pneumatic system, back- propagation (BP) algorithm is used to update the parameters of membership functions on-line. The partial derivative of the plant output with respect to the input, which is required by the learning process of FNN, is approximately estimated with a feed-forward neural network trained by recursive EKF. With the designed FNN controller for the pneumatic system, precise steady-state response and good dynamic tracking are obtained, which demonstrate that the nonlinear dynamics of the pneumatic system are efficiently overcome.