Reliability estimation from two types of accelerated testing considering individual difference and measurement error考虑个体差异与测量误差的两类加速试验可靠性评估
Yang CQ, Gu XH, Xiao ZM, 2024. Reliability estimation from two types of accelerated testing considering individual difference and measurement error. Probabilistic Engineering Mechanics, 75: 103584.DOI: 10.1016/j.probengmech.2024.103584
摘要 | Abstract
加速试验是一种提高试验效率和减少时间成本的有效方法,在可靠性试验中发挥着重要作用。本文介绍了一种可靠性评估的新方法,该方法同时有效利用加速退化试验 (accelerated degradation testing, ADT) 和加速寿命试验 (accelerated life testing, ALT) 数据,旨在全面利用产品的可靠性信息。为解决受污染退化数据和单元异质性带来的挑战,采用 Wiener 过程模型对产品退化过程进行建模,考虑个体差异和测量误差。该方法利用 Bayes 框架和 Markov 链蒙特卡罗算法,根据两种类型的加速试验数据进行参数估计。此外,模拟研究表明,与其它方法相比,该方法能够有效提高产品可靠性评估的准确性。对鱼雷引信橡胶 O 形圈的实际研究表明,该方法在解决实际问题方面具有很高的有效性。关键词: 可靠性评估, 加速试验, 个体差异, 测量误差, 橡胶 O 形圈Accelerated testing is a valuable approach for enhancing testing efficiency and reducing time costs, thus playing a crucial role in reliability testing. This paper introduces a novel reliability evaluation method that effectively utilizes both accelerated degradation testing (ADT) and accelerated life testing (ALT) data simultaneously, aiming to comprehensively leverage the reliability information of products. To address the challenges posed by contaminated degradation data and unit heterogeneity, the Wiener process is used to model the product degradation process, considering individual difference and measurement error. The proposed method utilizes a Bayesian framework with a MCMC algorithm for parameter estimation based on the two types of accelerated testing data. Furthermore, a simulation study shows that the proposed method can effectively improve the accuracy of product reliability estimation compared with other methods. A practical study for rubber O-rings of the torpedo fuse shows that the proposed method is highly effective in solving practical problems.Keywords: Reliability estimation; Accelerated testing; Individual difference; Measurement error; Rubber O-ringFig. 1. Flowchart of the proposed method
Fig. 2. Simulated degradation paths with measurement error
Fig. 3. Simulated measurement errors in ADT data
图 4: 本文所提方法、仅采用退化数据方法与忽略个体差异和观测误差方法的可靠性函数对比Fig. 4. Comparison of reliability functions with M0, M1, M2
图 5: 本文所提方法、忽略个体差异方法与忽略观测误差方法的可靠性函数对比Fig. 5. Comparison of reliability functions with M0, M3, M4
Fig. 6. Test fixture and sample
图 7: 50 °C, 60 °C, 70 °C, 80 °C 条件下的退化数据Fig. 7. Degradation data under conditions 50 °C, 60 °C, 70 °C, and 80 °C
Fig. 8. Reliability curves under four temperatures
Fig. 9. Reliability curve under 25 °C
Fig. 10. Lifetime distribution under 25 °C
作者信息 | Authors
杨承强 Cheng-Qiang Yang, 通讯作者 (Corresp.) 南京理工大学 (Nanjing University of Science & Technology) 机械工程学院Email: cq.yang1997@foxmail.com
南京理工大学 (Nanjing University of Science & Technology) 机械工程学院
新加坡南洋理工大学 (Nanyang Technological University) 机械与航空航天工程学院
律梦泽 M.Z. Lyu | 编辑 (Ed)
P.D. Spanos | 审校 (Rev)
陈建兵 J.B. Chen | 审校 (Rev)
彭勇波 Y.B. Peng | 审校 (Rev)