A degradation modeling and reliability estimation method based on Wiener process and evidential variable. (October 2020)

• Record Type: Journal Article
• Authors: Liu, Di; Wang, Shaoping
• Is Part Of: Reliability engineering & system safety. Volume 202(2020)
• Publication Date: 2020-10
• Subjects: Evidential variable -- Reliability estimation -- Wiener process -- Bayesian inference; 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.2020.106957
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: An evidential variable based reliability estimation method is proposed. The evidential variable is used to describe the model parameters. The proposed method can infer population performance without adding parameters. Bayesian inference method is used to evaluate and update the evidential variables. The proposed method is more accurate and stable than random variable based method. Abstract: Based on Wiener process and evidential variable, a reliability estimation and degradation modeling method is proposed in this paper. Wiener process is combined with evidence theory by applying evidential variable to describe model parameters. The basic probability assignments of evidential model parameters are evaluated based on likelihoods. Bayesian inference method is used to fuse degradation data and update evidential model parameters. Based on a wildly used actual degradation dataset, the distinguish features of the proposed method are verified. Due to powerful uncertainty describing ability of evidential variable, the proposed method shows superiority on online prediction and population evaluation compared to random variable based method, such as improvements in stability and accuracy. Furthermore, the proposed method can be used to evaluate population reliability without increasing parameter size, resulting in stochastic process become more practical under small sample conditions.