A Neural Network Scheme for Adaptive Control of Large Structures

点击翻译按钮获取中文摘要

In large flexible structures, control is required to eliminate vibrations induced by actions within and on the structure. The structure may comprise new materials which provide an inherent capability to actuate every strut in a truss or any discretized field of body forces in a solid. Furthermore, the structure configuration itself may change over time. Adaptive control is an appealing concept for such problems since it provides system identification and updating with system changes. This project applies adaptive control to an arbitrarily large structure, of the type previously described, and uses neural (massively parallel) computing to implement the control. Earlier neural network research suggest that a neural network approach may constitute a powerful architecture for the adaptive control of large structures. The research seeks fundamental analogues to prove or disprove a neural network capability for the large structures problem.