Time series methods in reliability

Abstract

The potential application of statistical time series methods for reliability analysis was first advocated by Nozer Singpurwalla in 1978. Apart from subsequent work by Singpurwalla and his co-workers, no other studies appear to have been published. Yet the established statistical time series techniques provide an exciting and promising approach for modelling empirical dependencies between successive times between failures. In this paper we consider the application of the widely used Box-Jenkins approach to explore and model the structures present in diverse reliability data. Other time series techniques, including Bayesian forecasting and multivariate time series methods, are discussed. We conclude that, whilst promising, the existing time series approaches require further specialisation in order to better model the characteristics of reliability problems.