An effective approach for kinematic reliability analysis of steering mechanisms. (December 2018)

• Record Type: Journal Article
• Authors: Wang, Lei; Zhang, Xufang; Zhou, Yangjunjian
• Is Part Of: Reliability engineering & system safety. Volume 180(2018)
• Page Start: 62
• Page End: 76
• Publication Date: 2018-12
• Subjects: Kriging surrogate models -- Kinematic reliability analysis -- Time-dependent reliability analysis -- Mechanisms; 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.2018.07.009
• Languages: English
• ISSNs: 0951-8320

Abstract

Highlights: The paper presents an effective Kriging model for kinematic reliability analysis. It standardizes various kinematic reliability problems via the extreme-value theory. The kinematic reliability method requires only one round of experimental analyses. Case studies have shown the proposed methodology has engineering applications. Abstract: The paper presents an effective surrogate model for system reliability analysis of mechanisms based on an extreme-value kinematic error model and the Kriging approximation. In this regard, two types of kinematic reliability problems are considered, i.e., the system reliability that requires the kinematic error remains within an accuracy threshold along an entire output trajectory, whereas the time-dependent reliability accounts for gradually propagated system failure domain due to the varied pseudo-time parameter from its beginning to the current realization. Both kinematic reliability problems need to evaluate the extreme-valued error function presented in the mixture of absolute and maximum operators. By contrast, the simple position-based error function allows developing accurate surrogate models with a rather small number of training samples. Therefore, an approximation model was proposed to deal with kinematic reliability problems of steering mechanisms. Compared to available reliability methods in the literature, results have demonstrated high accuracy of the proposed method for various kinematic reliability problems. Note that the proposed method for time-dependent reliability analysis requires only one round of design of experiment, rather than repeatedly running the kinematic model like the first-order reliability method. It therefore provides an effective simulation tool for kinematic reliability analysis of mechanisms.