Pavement Maintenance Planning at the Network Level with Principal Component Analysis

Abstract

Pavement management relies on the evaluation of the condition of pavement at different times during the life of the structure. The combination of condition indicators and the knowledge of how they are used in any pavement condition model is fundamental for pavement rating. To handle complexity and information redundancy, this paper proposes the principal-component analysis (PCA) to evaluate the relative importance of different types of distresses on the condition assessment for flexible pavements, and to use relevant condition features to establish criteria for pavement management. The important outcomes in applying the PCA approach were: even with the dimensionality reduction dictated by the variance, there was limited loss of information with regard to the section condition that did not affect the overall objective of pavement management; also, information redundancy was minimized.