Determination of degradation change point using spline function

Abstract

A degradation process is usually non-linear, and hence there exists one or more change points, where the degradation rate changes quickly. It is useful to determine the change points, e.g., designing an inspection scheme and predicting the time to a functional failure. It is an important issue how to specify the change points. This paper addresses this issue. A spline function with unknown knots is used to fit the observed degradation data and the estimated knots are the estimates of degradation change points. The information criterion is used to determine the optimal degree of spline function. The potential applications of the results are discussed, and the usefulness and appropriateness are illustrated by a real-world example.