Predicting Flush End‐Plate Connections Response Using Artificial Neural Networks

Abstract Predicting the moment‐rotation response parameters of semi‐rigid steel connections can be challenging given multitude components that contribute to connection's elastic and plastic deformations. This applies popular bolted flush end‐plate beam‐to‐column (FEPCs). The literature has highlighted limitations current analytical, mechanical, empirical models in providing accurate predictions. Considering these limitations, application machine‐learning methods structural engineering, such as artificial neural networks (ANN), have gained wide attention recently addressing problems associated with complex deformation damage phenomena. To end, superior nonlinearity ANNs is employed herein predict characteristics FEPCs. A dataset more than 200 specimens, collected from past experimental programs, utilized train ANN for predicting stiffness, strength, posy‐yield stiffness. paper describes deduction test data using fitting, determination significant geometric material features, architecture algorithms, accuracy metrics new model.