Optimal fault diagnosis strategy for complex systems considering common cause failure under epistemic uncertainty. Issue 9 (19th March 2021)

• Record Type: Journal Article
• Authors: Duan, Rongxing; Huang, Shujuan; He, Jiejun
• Is Part Of: Engineering computations. Volume 38:Issue 9(2021)
• Page Start: 3417
• Page End: 3437
• Publication Date: 2021-03-19
• Subjects: Fault diagnosis -- Epistemic uncertainty -- Common cause failure -- Dynamic evidence network -- VIKOR algorithm; Computer-aided engineering -- Periodicals; Computer graphics -- Periodicals; 620.00285
• DOI: 10.1108/EC-09-2020-0515
• Languages: English
• ISSNs: 0264-4401

Abstract

Abstract : Purpose: This paper aims to deal with the problems such as epistemic uncertainty, common cause failure (CCF) and dynamic fault behaviours that arise in complex systems and develop an effective fault diagnosis method to rapidly locate the fault when these systems fail. Design/methodology/approach: First, a dynamic fault tree model is established to capture the dynamic failure behaviours and linguistic term sets are used to obtain the failure rate of components in complex systems to deal with the epistemic uncertainty. Second, a β factor model is used to construct a dynamic evidence network model to handle CCF and some parameters obtained by reliability analysis are used to build the fault diagnosis decision table. Finally, an improved Vlsekriterijumska Optimizacija I Kompromisno Resenje algorithm is developed to obtain the optimal diagnosis sequence, which can locate the fault quickly, reduce the maintenance cost and improve the diagnosis efficiency. Findings: In this paper, a new optimal fault diagnosis strategy of complex systems considering CCF under epistemic uncertainty is presented based on reliability analysis. Dynamic evidence network is easy to carry out the quantitative analysis of dynamic fault tree. The proposed diagnosis algorithm can determine the optimal fault diagnosis sequence of complex systems and prove that CCF should not be ignored in fault diagnosis. Originality/value: The proposed method combines the reliability theory with multiple attribute decision-making methods to improve the diagnosis efficiency.