A Design of a Pneumatic Servo System Using a two-degree-of-freedom Controller and a Neural Network Compensator

In this paper, a combined control method using a two-degree-of-freedom controller and a neural network compensator is applied for trajectory control of a pneumatic cylinder. The two-degree-of-freedom controller consists of the I-PD controller,Smith's time delay compensator and disturbance compensator. The design procedure is essentially based on partial model matching which offers a reasonable way to simplify the design and controller configuration under disturbance rejection. In addition, a neural network compensator is used to compensate for modeling errors due to the non-linearity of the plant, and friction between cylinder and piston. Computer simulations and experiments of the trajectory control were carried out under various conditions. From the results summarized below, the effectiveness of the proposed control scheme could be confirmed :(1) Excellent reference tracking performance was obtained by adding the neural network compensator.(2) Excellent control performance for step disturbance was obtained.