Neural-model reference control of an air motor

Abstract

Neural networks are known for their unique properties of universal approximation. A description of a nonlinear plant in state space is obtained using a recurrent Elman network. Learning is implemented off-line with a truncated back propagation through time algorithm. Based on the neural model, a reference control strategy is implemented as a sequence of control actions computed by solving an optimisation problem iteratively. The past decade has seen an increase in research into the control of pneumatic drives. Two main reasons motivate this study: the response of pneumatic drives is slow, leading to an inability to attain set points due to high hysteresis; the dynamic model of the system is highly non linear, which greatly complicates controller design and development. Two streams of research efforts have evolved: using conventional methods to develop a modeling control strategy; adopting a strategy that does not require a mathematical model of the system. The paper presents investigations into the modeling and control of an air motor incorporating a pneumatic equivalent of the electric H-bridge. The pneumatic H-bridge has been devised for speed and direction control of the motor. The system is divided into three main regions: low speed (below 390 rev/min); medium speed (390 to 540 rev/min); high speed (540 to 680 rev/min). The system is highly nonlinear in the low speed region and hence an intelligent controller system, such as a neuro model and reference controller, is proposed