Building degradation index with variable selection for multivariate sensory data. (November 2022)

• Record Type: Journal Article
• Authors: Wang, Yueyao; Lee, I-Chen; Hong, Yili; Deng, Xinwei
• Is Part Of: Reliability engineering & system safety. Volume 227(2022)
• Publication Date: 2022-11
• Subjects: Adaptive LASSO -- General path model -- Prognostics -- Sensor selection -- Splines -- System health monitoring; 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.2022.108704
• Languages: English
• ISSNs: 0951-8320

Abstract

Abstract: The modeling and analysis of degradation data have been an active research area in reliability engineering for reliability assessment and system health management. As the sensor technology advances, multivariate sensory data are commonly collected for the underlying degradation process. However, most existing research on degradation modeling requires a univariate degradation index to be provided. Thus, constructing a degradation index for multivariate sensory data is a fundamental step in degradation modeling. In this paper, we propose a novel degradation index building method for multivariate sensory data with censoring. Based on an additive nonlinear model with variable selection, the proposed method can handle censored data, and can automatically select the informative sensor signals to be used in the degradation index. The penalized likelihood method with adaptive group penalty is developed for parameter estimation. We demonstrate that the proposed method outperforms existing methods via both simulation studies and analyses of the NASA jet engine sensor data. Highlights: A new framework to build degradation index based on cumulative exposure model. Use of both exact and censored data in model training for degradation index. Automatically select useful sensors to build the degradation index. Use splines to model the nonlinear effect of individual sensor signal. Simulation and data analysis show the proposed method outperforms existing methods.