1A1-F03 Construction of Bayesian network structure for fault diagnosis of event-driven systems : Evaluating importance of arc with Kullback-Leibler divergence(Manufacturing System and Manufacturing Machinery Mechatronics)

Abstract

This paper presents decentralized fault diagnosis strategy of event-driven systems based on probabilistic inference together with a method to construct the structure of the inference network, Bayesian network (BN). The structure of BN is essential since the computational burden and the fault diagnosis performance greatly depend on it. In this paper, we particularly focus on a construction methodology of BN structure based on evaluation of the arcs via Kullback-Leibler divergence of the conditional probabilities.