An Elman neural network approach in active control for building vibration under earthquake excitation

This article presents an improved Elman neural network for reducing building vibrations during earthquakes. The adjustment coefficient is proposed to be added to the Elman network's output layer to improve the controller's performance when used to minimize vibrations in buildings. The parameters of the proposed Elman neural network model are optimized using the Balancing Composite Motion Optimization algorithm. The effectiveness of the proposed method is assessed using a three-story structure with an active dampening mechanism on the first level. The study also takes into account two kinds of Elman neural network input variables: displacement and velocity data on the first floor, as well as displacement and velocity readings across all three floors. This research uses two measures of fitness functions in the optimal process, the structure's peak displacement and acceleration, to determine the best parameters for the proposed model. The effectiveness of the proposed method is demonstrated in restraining the vibration of the structure under a variety of earthquakes. Furthermore, the findings indicate that the proposed model maintains sustainability even when the maximum value of the actuator device is dropped.