A neural network is proposed in the paper to provide active control to a building-tuned mass damper system using direct control by a modified backpropagation scheme. The specialized learning architecture which allows the neural controller to learn to control the system interactively and autonomously while it is in operation is applied in this paper to overcome the effects of wind induced oscillations in tall civil structures. Simulation results are presented to study the properties of the proposed scheme and it is found that these results are encouraging for possible control implementations.
展开▼