Fast Response Estimation of Passively Controlled Structures Using Artificial Intelligence

The research on the applicability of using passive control systems as a seismic mitigation approach and energy dissipating devices has grown significantly in the past decades with the focus on investigating the seismic response of the structure equipped with viscous dampers. In general, viscous damper as a device can be utilized for retrofitting and rehabilitating of old buildings, in addition, to optimize the cost of the new ones. When used in a structure, these structural dissipating devices can enhance the energy absorption capacity of the building under shaking intensities resulting in decreased displacements and acceleration responses. Current analysis approaches for finding a safe and optimized design is considered to be time and effort consuming since it requires several trial and errors to achieve the targeted responses. Nowadays, the literature is full with many high reliable estimation methods, such as artificial neural networks, that can be used to define a mathematical prediction model based on certain input parameters to reduce the need for doing a long trial and errors. Therefore, the main objective of this study is to get benefit from these available estimation methods to propose a prediction models using ANN to obtain the response of structures equipped with viscous dampers from the behavior of the bare building. As a part of the study, thousands of numerical analyses by means of nonlinear response history analysis (direct integration method) are performed in order to estimate the response of various designs of bare and controlled reinforced concrete (RC) structures under different real earthquake records. Thereafter, the outputs of the numerical analyses will be used as an input for defining the coefficients of the estimation models using artificial neural networks (Levenberg-Marquardt backpropagation method) that will be proposed in this study. The proposed estimation models require one-fourth of the time to build the model and determine the results in comparison to NTHA which results in faster, minimized cost and effort approach. In addition to that, the proposed prediction models exhibited high performance and accurateness in estimating the responses of RC structures. Keywords: Reinforced concrete, optimization of dampers, damping, energy dissipation, artificial intelligence, estimation models, nonlinear response history analysis.