Use of Conceptual Graph-based Reasoning and Modeling to Support the Forensic Analysis of Bridges Failure

Abstract

Despite the development of mathematical models, procedures, decision support systems, and databases, bridges are still today confronted with failures that can lead to dramatic situations. The lack of tools allowing to learn from the past failures associated with the uncertainties in its environment makes difficult the decision making in the bridge maintenance. The difficulty of maintaining the functions of bridges comes from the complexity of relationships between random hazards, the limit states of the bridge structure, and the human decisions and activities. To date, there is no generic and holistic model capable of modeling all types of bridges failure situations and capitalizing on their knowledge. This paper proposes a knowledge model of bridge failures, based on the formalism of conceptual graphs, capable of representing and capitalizing the failures of all types of bridges in a structured common vocabulary and to conduct automatic reasoning. The aim is to be able to provide explanations of failures and alerts on future projects.