Reliability Analysis of Flow Meters with Multiple Failure Modes in the Process Industry
Reliability Analysis of Flow Meters with Multiple Failure Modes in the Process Industry
复制标题
DOI：
10.1109/rams48030.2020.9153604
复制
发表时间：
2020-01
期刊：
影响因子：
--
通讯作者：
Ying Liao;Yisha Xiang;Elias Keedy
中科院分区：
文献类型：
--
作者：
Ying Liao;Yisha Xiang;Elias Keedy
为了保证生产质量和运行安全，流量计的可靠性分析是流程工业企业面临的重要问题。在实际应用中，由于不同的机电部件产生多种失效模式，流量计失效过程的现场数据可能具有复杂的结构。此外，普遍存在不完整的记录。例如，安装日期通常不可用，导致故障数据左截断。也存在权利审查的情况，因为许多单位在分析数据时仍在使用。在这篇文章中，我们使用一个具有幂函数强度函数的非齐次泊松过程模型来处理左截尾和右截尾两种情况下现场数据的多种失效模式。利用极大似然法对模型参数进行估计。为了解决统计上的不确定性，采用随机加权似然Bootstrap方法估计参数的标准误差和可信区间。在案例研究中使用了来自流程工业公司的真实流量计故障数据。得到了强度估计函数和平均失效时间的估计，表明该参数模型能够很好地对失效数据进行合理的拟合。
Reliability analysis of flow meters is an important issue for process industry companies because of the need to ensure the production quality and operational safety. In practice, field data for flow meters failure process can have complicated structure due to multiple failure modes arising from different electromechanical parts. Besides, incomplete records generally exist. For example, the installation date is usually not available, making the failure data left-truncated. There also exist right-censored cases since many units are still in service when the data are analyzed. In this paper, we use a nonhomogeneous Poisson process model with power-law intensity functions to address multiple failure modes for field data with both left-truncated and right-censored cases. We apply the maximum likelihood method to estimate the model parameters. In order to address the statistical uncertainty, random weighted likelihood bootstrap procedure is used to estimate the standard errors and confidence intervals of the parameters. Real-world flow meters failure data from a process industry company are used in the case study. Estimated intensity functions and estimation of mean time to failure are obtained and show that the parametric model can reasonably fit the failure data well.