Exact Methods for Testing and Repairing a Failed Complex System with a Minimal Average Cost
Abstract
We consider any corrective maintenance operation of a complex system. The problem is to minimize the average cost of testing and repairing operations after each failure of the system under the assumption tests and repairs are perfect. We propose a series of methods for building optimal maintenance trees. Two improvement techniques are examined and experimented: use of memory and elimination with lower and upper bounds. || Nous considérons une opération de maintenance corrective d'un système complexe. Il s'agit de minimiser l'espérance de la somme des coûts de test et de réparation d'un système qui fonctionnait et qui vient de tomber en panne sous l'hypothèse que les tests