A Generic Framework for Generalized Virtual Age Models

This paper presents a generic framework for modeling and assessment of preventive and corrective maintenances. The virtual age (VA) models of Kijima are generalized. The proposed definition provide many existing models as VA models. New models are also proposed. Covariates effects can be considered. The framework is generic in the sense that we propose an iterative way to compute the characteristics of the model that does not depend on the number of maintenances, on their types, on their effect models, and on the way of planning the preventive maintenances. Such a framework is particularly interesting in order to develop adaptive software tools, such as the virtual age models (VAM) package of R language. Methods are proposed for maintenance times simulation, maximum likelihood parametric estimations, and reliability indicators computation. Finally, we present two applications to real dataset issued from off-road engines of Brazilian mining trucks and bladder cancer study.