Abstract
Pneumatic servo systems are used widely in industrial field. However, the dynamic characteristics often change with load mass and there are non-linear elements caused by air compressibility, various frictions and so on. Therefore, it is difficult for conventional linear control methods to accomplish satisfactory control performance. In addition, there is a problem that discrete-time models of pneumatic servo system have possibilities to become non-minimum phase by difference of the sampling period, change in load mass, and so on.
In this paper, for these problems, we propose a design scheme which combines a Model Reference Adaptive Control (MRAC) using delta-operator and a Neural Network (NN). In this design, even if the discrete-time model is non-minimum phase, it can be constructed by regarding approximate model of the plant expressed by delta-operator as minimum-phase model, and NN compensates for modeling error and the non-linearity of pneumatic servo systems.
