Development of a Bridge Performance Prediction Model as a Rational Basis for a Structures Management System.

摘要

Bridge management is one component of the transportation and pavement management infrastructure. Like pavement management, bridge management is a large-scale complex system, but can be treated separately, since bridges exude some independent characteristics. Throughout its operational life, bridges as do pavements require routine and periodic maintenance and rehabilitation. However, a bridge lasts much longer than a pavement section. For a highway agency, bridges are a long-term multi-year investment. Separate bridge management analysis is, therefore, desirable and essential with these future economic decisions in mind. The primary objective of the project is the development of an appropriate performance model that will accurately predict how rapidly and in what fashion bridges deteriorate. The predictive capability of the model will then be incorporated into a bridge management system framework for the consideration of policy management and expenditure optimization issues. A Markovian Decision Process (MDP) is a probabilistic analysis approach that is being utilized in the project. The MDP is a powerful tool for formulating models and finding optimal policies for controlling a large-scale system. The complexity combined with the dynamics and predictive needs of the large-scale bridge management system create an appropriate application for MDP use.