详情

原文传递 ColumnsNet: Neural Network Model for Constructing Interaction Diagrams and Slenderness Limit for FRP-RC Columns

题名：	ColumnsNet: Neural Network Model for Constructing Interaction Diagrams and Slenderness Limit for FRP-RC Columns
正文语种：	eng
作者：	Ahmad Tarawneh;Ghassan Almasabha;Yasmin Murad
作者单位：	The Hashemite Univ.;The Hashemite Univ.;Univ. of Jordan
关键词：	Fiber-reinforced polymer (FRP)-RC;Artificial neural network (ANN);Machine learning;Slenderness limit;Interaction diagrams;FRP bars
摘要：	Abstract Predicting the axial capacity and behavior of concentrically, eccentrically, and slender loaded fiber-reinforced polymer (FRP)-RC columns is not completely established, and the current design codes lack design provisions for FRP-RC columns. Rather, it requires ignoring the contribution of FRP bars in compression conservatively. To bridge this knowledge gap, this study proposes an artificial neural network (ANN)-based model capable of predicting the axial capacity and slenderness limit and constructing an interaction diagram for FRP-reinforced columns. The aforementioned model was trained with Bayesian regularization utilizing a comprehensive database of 241 tested FRP-RC columns. Parameters included in the model are column cross-sectional area, compressive strength, FRP elastic modulus, reinforcement ratio, eccentricity ratio, and slenderness ratio. The predictions of the ANN-based model match well with the experimental results of the compiled database; the model predictions have a COV of 15% and root-mean square error of 130 kN. In addition, a parametric study was conducted to investigate the effect of parameters and ensure the generalizability of the proposed model.
出版年：	2022
期刊名称：	Journal of structural engineering
卷：	148
期：	8
页码：	04022089.1-04022089.12