Integrated importance measures of multi-state systems under uncertainty

• Published in: Computers & industrial engineering Vol. 59; no. 4; pp. 921 - 928
• Main Authors: Si, Shubin, Cai, Zhiqiang, Sun, Shudong, Zhang, Shenggui
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 01.11.2010; Pergamon Press Inc
• Subjects: Algorithms; Bayesian analysis; Bayesian network; Causality; Construction; Criteria; Head-up display; Industrial engineering; Integrated importance measure; Mathematical analysis; Multi-state system; Networks; Reliability engineering; Studies; Systems analysis; Uncertainty; Uncertainty; Multi-state system; Head-up display; Integrated importance measure; Bayesian network
• ISSN: 0360-8352; 1879-0550
• DOI: 10.1016/j.cie.2010.09.002

Importance analysis in reliability engineering is used to find the weakest components in a system. Traditional importance measures for multi-state systems analysis mainly pay attention to the reliability or structure characteristics of components, but seldom consider the causalities between components in the system under uncertainty. In order to solve the above problems, the multi-state system Bayesian network is proposed to represent the multi-state system under uncertainty and facilitate the component importance calculation. Then, this paper puts forward a separate subset algorithm based on the Bayesian information criterion and K2 algorithm to build the multi-state system Bayesian network of practical systems automatically. By considering the reliability, structure and causality characteristics of components comprehensively, the integrated importance measure is also presented to describe the effects of component failures on the state distribution of the multi-state system under uncertainty. Finally, the application of the multi-state system Bayesian network, the separate subset algorithm and the integrated importance measure in a simple head-up display system is implemented to verify the effectiveness of the proposed methods in components importance analysis.