An approach for reliability prediction of instrumentation & control cables by artificial neural networks and Weibull theory for probabilistic safety assessment of NPPs. (February 2018)

• Record Type: Journal Article
• Authors: Santhosh, T.V.; Gopika, V.; Ghosh, A.K.; Fernandes, B.G.
• Is Part Of: Reliability engineering & system safety. Volume 170(2018)
• Page Start: 31
• Page End: 44
• Publication Date: 2018-02
• Subjects: Artificial neural networks -- Insulation resistance -- Accelerated life testing -- Weibull reliability -- Probabilistic safety assessment -- Nuclear power plants; 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.2017.10.010
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: An approach for reliability prediction of I&C cables based on ANNs is proposed. Prediction of time-dependent reliabilities for use in PSA of NPPs is demonstrated. Comparison of alternate models with ANN has been discussed. The issue of modeling synergistic effects in reliability analysis is addressed. Statistical significance tests performed suggest good confidence in ANN findings. Abstract: The polymeric materials used for insulation and sheath in instrumentation and control (I&C) cables of nuclear power plants (NPPs) are subjected to degradation due to various stressors. The prediction of long-term aging and lifetime of cables is generally determined based on accelerated life testing (ALT) experiments which are not only expensive but also time consuming. Application of artificial neural networks (ANNs) in the field of transient diagnosis and condition assessment of electrical and other equipment has been a promising technique; however the use of ANN for reliability prediction of I&C cables has not yet been studied. This paper presents an integrated approach to predict the lifetime and reliability of I&C cables by ANN from the accelerated aging data. In order to validate the proposed methodology for use in probabilistic safety assessment (PSA) of NPP to account for the cable failures, ALT data on a typical cross-linked polyethylene (XLPE) insulated I&C cable has been referred from the literature. The time dependent reliability was predicted by considering the various failure rates. Study demonstrates that by an appropriate training algorithm with suitable network architecture, it is possible to predict the reliability of I&C cables by ANN with the minimal accelerated life testing.