Analyzing accelerated degradation data by nonparametric regression

Abstract

This paper presents a nonparametric regression accelerated life-stress (NPRALS) model for accelerated degradation data wherein the data consist of groups of degrading curve data. In contrast to the usual parametric modeling, a nonparametric regression model relaxes assumptions on the form of the regression functions and lets data speak for themselves in searching for a suitable model for data. NPRALS assumes that various stress levels affect only the degradation rate, but not the shape of the degradation curve. An algorithm is presented for estimating the components of NPRALS. By investigating the relationship between the acceleration factors and the stress levels, the mean time to failure estimate of the product under the usual use condition is obtained. The procedure is applied to a set of data obtained from an accelerated degradation test for a light emitting diode product. The results look very promising. The performance of NPRALS is further checked by a simulated example and found satisfactory. We anticipate that NPRALS can be applied to other applications as well.