Bayesian Network Methods for Modeling and Reliability Assessment of Infrastructure Systems

Abstract

Infrastructure systems are essential for a functioning society. These systems, however, are aging and subject to hazards of increasing frequency and severity. This chapter presents novel Bayesian network (BN) methodologies to model and assess the reliability of complex infrastructure systems. BNs are particularly well suited to the analysis of civil infrastructures, where information about the systems is often uncertain and evolving in time. In this environment, BNs handle information probabilistically to support engineering decision making under uncertainty, and are capable of updating to account for new information as it becomes available. This chapter addresses one of the major limitations of the BN framework in analyzing infrastructure systems, namely the exponentially increasing memory storage required as the size and complexity of the system increases. Traditionally, this has limited the size of the systems that can be tractably modeled as BNs. Novel compression and inference algorithms are presented to address this memory storage challenge. These are combined with several heuristics to improve the computational efficiency of the algorithms. Through the application of these algorithms and heuristics to example systems, the proposed methodologies are shown to achieve significant gains in both memory storage and computation time. Together, these algorithms enable larger infrastructure systems to be modeled as BNs for system reliability analysis.