摘要: |
A bridge management system is designed to provide information and useful data for analysis so that bridge engineers can make more consistent, cost-effective, and defensible decisions for the maintenance, rehabilitation, and preservation of bridges. In a bridge management system, problem setting often involves many uncertain, interrelated quantities, attributes, and alternatives derived from information of highly varying quality. Recently, there has been considerable interest in addressing uncertainties in bridge management. This paper proposes the use of a type of Bayesian influence diagram known as a directed acyclic graph (DAG). DAGs express outcomes in terms of combinations of primitive events. The graphical structure of these models also captures the dependency structure among events, which enables the decision maker to exploit conditional independence to reduce specification and computation. The influence diagrams give users a clear view of the variables within the framework of a bridge management system and the relationships among them. Influence diagrams provide an appealing knowledge representation that can express uncertain knowledge, beliefs, and preferences in both qualitative and quantitative forms. With influence diagrams, the question can be addressed of perfect and imperfect information in bridge management system decision making. |