Optimum connection of LSF system braces using the seismic-ANN approach

Abstract

Lightweight steel framing (LSF) has been proposed as an economic and earthquake-resistant system. The tendency of mass constructors to use this system is due to it being a full industrial process. One of the systems that resist lateral load in cold-formed steel structures is the application of braces. Optimization and improvement of connections for these braces have been considered by experts in this field of research. In this paper, experimental studies and normalization and simulation by artificial neural network (ANN) were used. The results of the research have been applied to create a nonlinear relationship. All input and target data must be normalized and then simulation and training by a neural network can be performed. In this research, two layers have been used. One of these is a sigmoid layer. The results show that optimal connections in light weight steel framing systems have suitable plasticity, load capacity and nonlinear relations. Statistical analysis with SPSS software shows that there is no significant difference between the neural network and experimental results (P-Value > 0.05).