Building degradation index with variable selection for multivariate sensory data

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