A discrete-time Bayesian network approach for reliability analysis of dynamic systems with common cause failures. (December 2021)

• Record Type: Journal Article
• Authors: Guo, Yongjin; Zhong, Mingjun; Gao, Chao; Wang, Hongdong; Liang, Xiaofeng; Yi, Hong
• Is Part Of: Reliability engineering & system safety. Volume 216(2021)
• Publication Date: 2021-12
• Subjects: Dynamic system -- Reliability modelling -- Discrete-time Bayesian network -- Common cause failure -- Diagnosis; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2021.108028
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: Reliability analysis model for dynamic systems with CCF based on DTBN. DTBN models are developed for a parallel system with cold and warm spare parts. Logical relationship between failure states is examined to obtain CPT of DTBN node. Impact of CCFs on a digital safety-level DCS of NPPs is studied by proposed method. Abstract: The dynamic and dependant behaviors are typical characteristics of modern complex systems, whose reliability is often improved through the design of multichannel parallel structures. The existence of common cause failures (CCFs) has a significant impact on system reliability. A reliability analysis model is proposed for dynamic systems with CCFs based on discrete-time Bayesian networks (DTBNs). The system operating time is dispersed into several time intervals, and the component failures are divided into independent and CCF states. Dynamic systems with cold and warm spare parts are examined to determine the modelling methodology and conditional probability tables (CPTs) of Bayesian network (BN) nodes. The reliability calculation is realised through the Bayesian inference mechanism. The model is applied to the CCF analysis and fault diagnosis of a digital safety-level distributed control system (DCS) of nuclear power plants (NPPs) to prove the effectiveness and feasibility of the method.