Model-based predictive digital twin for bridge structural health monitoring: the integration of building information modeling, finite element analysis, and machine learning in a Netherlands case study

Advancements in bridge health monitoring systems are a trending issue for many researchers and industry professionals these days, where the bridge community is looking for Artificial Intelligence (AI) based automated solutions. To address this challenge, this study integrates Building Information Modeling (BIM), Finite Element Method (FEM), and Machine Learning (ML) algorithms to create a comprehensive Digital Twin (DT) framework capable of real-time monitoring and predictive analysis of the bridge's structural integrity. A robust design of the workflows and technical framework is proposed to enable seamless interaction between the physical bridge assets and the proposed model-based Predictive Digital Twin (PDT) system using virtual designs and Neural Network (NN) prediction algorithms. The major research contributions of this article include model-based Bridge Structural Health Monitoring (BSHM) using accurate predictions from ML and sensor data and the novel PDT-based approach. It proposes information flows through an integrated BIM, ML, and DT platform to ensure improved BSHM. The proposed ML-based PDT framework is validated using a steel girder bridge located in the Netherlands, which demonstrates automated health monitoring of the bridge. The PDT system is validated using actual BSHM data (strain data), with predictions from the ML model compared against known load cases. The outcomes of this research showcase the PDT framework's ability to predict load placement and magnitude with precision, contributing to the ongoing advancement of suitable infrastructure management practices.