Integrating multivariate techniques in bridge management systems for life-cycle prediction

Abstract

The use of bridge management systems (BMS) by infrastructure stakeholders has led to the collection and retention of large quantities of data concerning the condition states of bridges throughout national and regional networks. The database for the BMS is often populated by the results of routine visual inspections, based on a prescribed scale for defining the condition state of the bridge’s individual elements, and of the bridge structure as a whole. The populating of the database also leads to the storage of large quantities of so-called metadata; which can describe the physical parameters of the bridge. The availability of this data allows the assessment of the BMS using multivariate techniques to enhance the life-cycle assessment of bridge networks, through advanced descriptive and predictive techniques applied to deteriorating network assets. Multivariate techniques such as principal component analysis have been demonstrated by the authors to be effectively applied as a descriptive tool to an existing BMS, and the results of a case study of a large dataset of bridges indicate its viability to be integrated into data-based approaches to infrastructural asset management.