Premium
Component lifetime models based on weibull mixtures and competing risks
Author(s) -
Hansen Christian Kornerup,
Thyregod Poul
Publication year - 1992
Publication title -
quality and reliability engineering international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.913
H-Index - 62
eISSN - 1099-1638
pISSN - 0748-8017
DOI - 10.1002/qre.4680080405
Subject(s) - weibull distribution , exponential function , component (thermodynamics) , class (philosophy) , hazard , model selection , mathematics , statistical model , bathtub , computer science , statistics , artificial intelligence , history , mathematical analysis , physics , chemistry , organic chemistry , archaeology , thermodynamics
This paper gives a review of the statistical modelling of lifetimes of electronic components. The models considered are all based on Weibull (and exponential) distributions combined as either finite mixtures, competing risks or both. A convenient notation is introduced for this general class of models, which includes the traditional monotone hazard rate models, the bathtub model, as well as the more sophisticated roller coaster models. The main advantage of this concept is that a model may be determined by relatively few parameters, and that furthermore each parameter may have a physical interpretation in terms of initial flaws or defects as well as long‐term wear out or deterioration. The properties of the distributions are investigated in examples with different scale and shape parameters, and for different types of combinations. Also, the models are compared with another class of models based on combinations of exponential distributions, known as the phase‐type models. Finally a brief discussion of model selection and estimation methods is included.