Prediction of Nonlinear Time-History Inter-story Drifts of Planar Steel Moment Frames Using Neural Network Techniques

Inter-story drifts are important indicators of the frame performance and damage level of planar S.M.R.F. under earthshaking excitations. However, predicting them accurately is challenging due to the complex non-linear behavior of steel frames and the uncertainty of Earthshaking Earth vibrations. The Complicated behavior method can precisely compute the frame response to intense earthshaking shaking when applied correctly. N.L.T.H.A is a method of analyzing the dynamic response of a Framework under a time-varying load. The Complicated behavior Analyzing model encompasses the inelastic behavior of frame members during cyclic earthshaking Earth vibrations. As a result, the Complicated behavior method directly simulates the dissipation of hysteretic energy within the non-linear region of the Framework. However, this method is computationally expensive and requires Numerous ground motion records to obtain reliable results. This paper explores the application of neural network training for the Anticipation of inter-story drifts of special S.M.R.F. Frames under earthshaking loading, as an alternative to the historical record Analyzing method. Neural network training is a method that uses artificial neural networks (ANNs) to learn from data and make Anticipations or classifications. A data set of ۲۰۰ and ۵۰۰ samples is randomly created to train and test the NN models for predicting the non-linear historical record inter-story drift ratios of a ۶- and ۱۲-story special steel support structure. The results show that the CFBP NN model with ۱۵ hidden layer neurons is better than the other models for predicting the inter-story drift ratios at performance levels for both the ۱۲ and ۶-story steel support structures.

Keywords:

Outcome-Driven Design , Steel support structure , Neural Network , Historical record Analyzing