Addressing Uncertainties in Flexible Pavement Maintenance Decisions at Project Level Using Bayesian Influence Diagrams

by Busby Attoh-Okine, Univ of Kansas, Lawrence, United States,



Document Type: Proceeding Paper

Part of: Infrastructure: Planning and Management

Abstract: Bayesian influence diagrams are directed acyclic graphs used as a network representation for making decisions under uncertainty. These representation facilitate the assessment of coherent pair distributions and make it easier for knowledge engineers and experts to express and understand more general kinds of dependence and independence assumptions. As well as having rigorous formal interpretation, influence diagram have well quantitative structure that make it appropriate to show causality and flow of information. Bayesian influence diagrams offer an important complement to more traditional representations such as decision-analytic frame-work for reasoning about flexible pavement maintenance at the project level. The influence diagram connects performance-related factors, historically-related factors of pavement, policy related factors in maintenance, geometrically-related factors of the pavement, environment and cost related issues. The following points are highlighted: (a) Identifying the key points of a pavement maintenance decision problem, (b) establishing the guidelines for quality decision making in pavement maintenance, (c) quantifying uncertainties in pavement maintenance, (d) assessing the guidelines for quality engineers and policy makers, (e) determining the best maintenance decisions, and (f) estimating the value of information during flexible pavement maintenance at the project level. The approach appears straight-forward and practical in structuring and constructing pavement maintenance and analyzing pavement maintenance problems under a variety of prevailing risk attitudes.

Subject Headings: Maintenance | Asphalt pavements | Uncertainty principles | Bayesian analysis | Information management | Quality control | Risk management

Services: Buy this book/Buy this article

 

Return to search