Cee 249 Seminar

Ms. Iris Tien, PhD Candidate, UC Berkeley
Bayesian network methods for modeling and reliability assessment of infrastructure systems

Ms. Iris Tien, PhD Candidate, UC Berkeley

Bayesian network methods for modeling and reliability
assessment of infrastructure systems


Abstract

The Bayesian network (BN) is an ideal tool for modeling and
assessing the reliability of civil infrastructure, particularly when
information about the system and its components is uncertain and evolves in
time. The major obstacle to the widespread use of BNs for system reliability analysis,
however, is the limited size and complexity of the system that can be tractably
modeled as a BN. This is due to the exponentially
increasing number of elements that must be stored in the conditional
probability table (CPT) for the system node in the BN as the number of
components in the system increases. In this seminar, I will present novel compression
and inference algorithms that I have developed to address this limitation. The
algorithms utilize compression techniques to achieve orders of magnitude savings
in memory storage for the system CPT. In addition, heuristics developed to
improve the computational efficiency of the algorithms are presented. The
algorithms are applied to the analysis of an example system, and the
performance evaluated compared to existing methods. The gains achieved by the
developed algorithms in both memory storage and computation time are
demonstrated.