Data-driven life-cycle risk assessment of bridge networks using Bayesian network

Abstract

Bridge failures within the transportation network can lead to significant losses. Such failures are attributable to many factors from design to usage, internal to external, historic to present. It is therefore important to monitor and assess these risks in a holistic life-cycle approach. Generally, life-cycle risk assessment involves the evaluation of bridge failure probability over time and the associated consequences/losses. With increasing capabilities to harvest big data from various sources, the risk profile can be updated by integrating the power of data into the assessment. This paper presents a framework based on Bayesian network to perform data-driven life-cycle risk assessment for bridge networks. The life-cycle risk is updated with data from count stations and the bridge monitoring data. 13 bridges in a bridge network in New Jersey is used to demonstrate how the framework fuses the data and complex model of the system for risk assessment of the bridge network.