An Improved Algorithm of Structure Learning Applied in Organizational Factors Bayesian Belief Network

摘要

The paper first introduces the concept of organizational factors in the socio-technical system and the basic theory of Bayesian Network and then discusses the algorithm of structure learning of Bayesian network. A new algorithm base on dependency analysis is proposed to effectively reduce the number of detecting condition independence. It uses heuristic cutset searching algorithm and orients all the edges in the network before removing superfluous edges. Experiment results indicate that it outperforms the traditional algorithm. Finally, organizational factors Bayesian network is constructed by using the algorithm which is helpful for the nuclear pant to discover the critical factors influencing human reliability in nuclear power plant.