Neural Network for Estimation of Nonlinear Structure Dynamics

Developing robust adaptive control strategies for realworld civil structures has been a topic of recent interest. The motivation for exploring adaptive techniques comes from the acknowledgment that since structures behave in unexpected forms and non-linearly when excited by strong-ground motions, the implementation of conventional xed controller strategies may prove to be naive. Often the governing response properties only exhibit themselves for the rst time when subjected to strong shaking. As a result of this, control strategies should incorporate exible adaptive identi cation schemes which can quickly capture and emulate the essential response signature of a structural system and react accordingly. Of course, another key feature of adaptive techniques is that they can model timevarying behavior, for example, structural deterioration is often observed during the course of strong ground excitation.