Reconfigurable learning control in large space structures

The design of control algorithms for large space structures, possessing nonlinear dynamics which are often time-varying and likely ill-modeled, presents great challenges for all current methodologies. In the present paper, we propose the use of a hybrid connectionist system as a learning controller with reconfiguration capability. The ability of connectionist systems to approximate arbitrary continuous functions lends itself as an efficient controlling mechanism. A fault diagnosis network is applied for health monitoring to provide the neural controller various failure scenarios. Associative memory is incorporated into an adaptive architecture to compensate slowly-varying as well as catastrophic changes of structural parameters by providing a continuous solution space of acceptable controller configurations. The paper outlines the theoretical foundation of the architecture and demonstrates its applicability via specific example