Enhancing seismic performance prediction of RC frames using MFF-ANN model approach

The seismic response of human-made structures to ground shaking caused by earthquakes can lead to catastrophic damage. Seismic investigation, a sub-discipline of primary examination, is utilized to evaluate the seismic reaction of designs. Artificial intelligence has emerged as a solution to address this problem. The seismic response of reinforced concrete (RC) is investigated in this research using an artificial neural network (ANN) structures to ground motions. The evaluation of a structure's seismic response is crucial for upgrading a building or its components. As a novelty of this study, Extended three-dimensional analysis of building systems (ETABS) is used to determine the seismic response of all structures, which serves as target data for designing the ANN. Symmetrical buildings are stimulated using various ground motions, and the resulting input and target data are used to construct an ANN in MATLAB. A novel multi feed-forward type of ANN (MFF-ANN) with the Levenberg Marquedt algorithm is employed. The input parameters that produce the lowest error and highest accuracy for forecasting the seismic response of RC multistory buildings are identified. The significance of each parameter used in the input layer contributing to the maximum accuracy is determined, along with the percentage of each parameter contributing to the optimal network.