Artificial Intelligence for Structural Reliability Analysis of Steel Truss Bridges

Artificial Intelligence (AI) has become an essential technology in civil engineering, enabling innovative approaches to solving complex problems, optimizing decision-making processes, and enhancing efficiency in design, analysis, and construction management. Within this scope, the primary objective of this paper is to evaluate the potential of Artificial Neural Networks (ANN) as an efficient predictive model for analyzing the structural response of steel Pratt truss bridges, specifically within the context of reliability analysis. A conceptual Pratt bridge configuration with twelve panels is considered alongside structural data generated from 84 cases under varying geometric and loading conditions. The ANN model is trained to predict the maximum tensile and compressive forces in key truss elements, including diagonal, lower, vertical, and upper chords, using input parameters such as truss geometry and applied forces. In addition, while comparing the ANN predictions against classical methods, the study further employs Monte Carlo Simulation (MCS) to assess the probability of failure and reliability index of the most loaded structural elements. The study demonstrates the ANN's capability to deliver accurate predictions with reduced computational costs, highlighting its potential for integration into advanced structural analysis frameworks.