A dynamic Bayesian networks-based approach for data fusion of model-based SHM and visual inspections in post-tensioned bridges

Bridges are primary infrastructure assets that facilitate human mobility and drive economic and social activities. The integrity, safety, and serviceability of bridges are essential requirements that must be maintained throughout their lifespan. In this context, Structural Health Monitoring (SHM) has emerged as a useful tool for addressing challenges related to aging infrastructure management, providing real-time information on bridge behavior. On the other hand, visual inspections remain essential for identifying visible defects, but they are often not sufficient to provide a complete picture of a bridge’s health condition. Hence, integrating the information from both SHM systems and visual inspections is essential to maximize their combined potential. To accomplish this challenge, Bayesian networks (BN) are a competitive solution for handling uncertain and incomplete information. A BN is a graphical model that uses Bayesian inference for probability computations. Each node in the network represents a random variable, and each edge represents the conditional probability of one random variable given its parents. The extended version of Bayesian networks in the time dimension is designated as a Dynamic Bayesian Network (DBN). This paper presents a modular framework that fuses time-dependent data from SHM and visual inspections using a DBN, applied to selected existing bridges. The framework is generally applicable across different bridges and damage scenarios without requiring a data-driven training phase. This approach demonstrates the potential for timely interventions when anomalies arise, enhances ongoing knowledge about each bridge’s condition, and facilitates strategic planning for both short- and long-term maintenance. By consolidating all stored data, the framework aims to optimize resource allocation and support economically efficient infrastructure management.