ABSTRACT
The task of decision support for optimal management of infrastructure systems, including bridges, is a non-trivial one. The advent of measurement and monitoring technologies allows for conveying a wealth of diversified information, including measurements on structural condition, environmental conditions (temperature, humidity, rainfall), geo-information (earth movement), hazards data (e.g. earthquake data), as well as operating loads (e.g. traffic). Despite the generated Big Data stream, mere existence of this data in raw form is not meaningful for decision support. Instead, the initial data gathering, which is the task of sensing, needs to be translated into meaningful indicators of performance and condition, or interpreted into understandable events, which pertains to the task of sense-making. As part of Task 4.4 of the FORESEE project, we have delivered a diagnostic and prognostic framework for infrastructure decision support that capitalizes on data extracted from models (simulations) as well as monitoring information. We define an object-oriented framework, relying on graphical model structures (Random Forests and Bayesian Networks), which considers monitored structures as “systems of systems”, with each system comprising an assembly of objects. We demonstrate the decision support framework on the illustrative examples of a monitored bridge and a roadway network.
