Identification and control of large structures using neural networks

Abstract

Control system design for large space structures, possessing nonlinear dynamics which are often time-varying and likely ill-modeled, presents great challenges for all currently advocated methodologies. The pursuits of an autonomous control system for such nonlinear structures have led to the use of artificial neural networks. In the present paper, we propose the use of radial basis function networks as a learning controller to achieve vibration suppression and trajectory maneuvering. The ability of connectionist systems to approximate arbitrary continuous functions provides an efficient means of modeling, identification and control of complex systems. Based on the model reference adaptive control architecture, a neural controller learns to function as a closed-loop compensator and to force the dynamics of the nonlinear plant to match a given reference model. This paper addresses the theoretical foundation of the architecture and demonstrates its applicability via several examples.