Artificial Intelligence-Based Computational Analysis and Optimization of Civil Engineering Structures

This paper examines the possibility of artificial intelligence (AI) approaches for the computational representation of civil engineering structures, with specific attention paid to prediction of reinforced concrete beam capacity. A comparative study of Deep Neural Networks (DNN) and traditional Finite Element Analysis (FEM), Support Vector Machines (SVM) in dimension jump for computational efficiency and predictive accuracy The experimental results show that DNN is the best choice for resolving such complicated nonlinear problems as well as multi-dimensional features, showing the R-2 and MSE of 0.92 and 0.025 respectively, while the computation time is just about only 1/10 compared to FEM. Although AI models present remarkable efficiency and accuracy, drawbacks like data dependency and lack of model interpretability are still there. This paper presents the highest level of effort in improving data quality, optimization on feature selection, and a more general approach to exposing AI-based models embedded inside the structural computational model with accompanying physics-informed AI that can strengthen its application in structural computation. Future research will focus on multi-source data integration, advanced model optimization, and intelligent engineering applications, contributing to the digital transformation of civil engineering.