| Abstract |
: |
The maintenance of complex systems is a discipline that requires great knowledge and great experience. These factors are, unfortunately, not enough: the opportunity to model the system to be maintained and to represent it in the form of a mathematical and probabilistic model, can’t be disregarded at all. These composite systems, moreover, are often subject to problems of uncertainty and must operate in multifunctional contexts. So, even the modelling tools must be able to take into account these aspects. In recent years, new probabilistic tools, such as the Dynamic Object oriented Bayesian Networks (DOOBNs), have been developed. They’re able, among other things, to estimate the reliability performance of complex systems and have been applied in production plants, machinery and process plants. An usual problem, often occurring, is to correctly represent the process of imperfect maintenance. This is characterized by a recovery of the system that brings it back into an intermediate situation between the minimum repair and perfect repair. The aim of this study is to evaluate how it is possible to use dynamic Bayesian networks to model imperfect maintenance. In particular, first we studied the problem, then a case study on the compressed air production and treatment system of an high speed train was analysed and finally the analytical capabilities of the instrument were assessed. The originality of this work lies in the fact that there are not in the literature DOOBN applications to model imperfect maintenance. The study results were very satisfactory and in the future it will be necessary to better investigate what are the limits and the potentials of the instrument. |
