An Artificial Neural Network Approach to Pavement Maintenance Decision Support System

by Abdullah M. Alsugair, (A.M.ASCE), King Saud Univ, Riyadh, Saudi Arabia,
Essam Sharaf, King Saud Univ, Riyadh, Saudi Arabia,

Document Type: Proceeding Paper

Part of: Computing in Civil Engineering


An important component of any Pavement Maintenance Decision Support System (PMDSS) is the condition survey and rating procedures. Data obtained from these procedures are the primary basis for determining Maintenance and Repair (M&R) actions. These actions are typically determined through the use of prescribed intervention logic (rules) which recommends maintenance actions on the basis of current pavement condition. This paper investigates the feasibility of using Artificial Neural Network (ANN) to recommend appropriate M&R actions. In order for an ANN to diagnose an M&R action accurately, it must be trained with correct diagnosed M&R actions (training sets). Each training set consists of pavement condition and the corresponding recommended M&R action. In this study, pavement condition data used in the training sets were obtained from a comprehensive visual inspection data conducted on the Egyptian Road Network. The associated M&R actions were obtained based on consulting human expertise (district managers and engineers) as well as M&R actions recommended by the PAVER (a pavement management system developed by the U.S. Army Corps of Engineering). The results of this study reveal that ANN, trained using pavement conditions and M&R actions, has a strong potential for implementation in the PMDSS.

Subject Headings: Pavement condition | Neural networks | Maintenance and operation | Decision support systems | Pavement overlays | Training | Pavements | Ratings

Services: Buy this book/Buy this article


Return to search