Abstract
The base isolation is a promising technique to eliminate the damage from ground motion. In order to improve the safety of a semi-active building, the magneto-rheological (MR) dampers are used to replace the traditional adjustable dampers. As the MR damper features nonlinearity, it is not easy, and even impossible, to apply the traditional control strategies on the basis of precise mathematical models. Hence, this paper presents an indirect adaptive control structure based on neural networks. This control structure consists of a neural controller, and a neural identifier which can estimate the back-propagation error of controller. The numerical simulations show that the new strategy is much better than the results of neural network controller without quadratic momentum term.
