A Model for Continuous Degradation Data.

In this paper, a model for continuous degradation data is introduced. The model is based on the fact that observable degradation rates for processes which can be modeled by a linear system of first order differential equations (DE's) have a linear structure themselves. Since many dynamic systems can be approximated by a linear system of first order DE's, this model has some potential of being applicable in a large variety of situations. Once the model is derived and estimation procedures developed, it is applied to the analysis of data from a small step stress experiment conducted on InGaAsP lasers. This application demonstrates that: i) The modelling procedure yields reasonable results when applied to real degradation data (predicting ~ 90% of the variation in the degradation rates at operating conditions). ii) The modelling procedure has the potential to greatly reduce the cost and time required for certifying components for high reliability work. In addition, the application emphasizes some areas, such as statistical inference and error estimation procedures, which require future research. Application to other structures, such as integrated circuits, connectors, and printed wiring boards, will provide further insights.