Abstract:
This thesis focuses on optimizing the frequency of preventive maintenance (PM) of complex
repairable systems. This optimization requires establishing a consistent reliability model
capable of perfectly describing the system’s behavior and taking into account the influence of
the different covariates on the system’s operation. An application on a history of three
turbochargers, working in an oil industry, allowed us to consider the temperature, the
programming of maintenance actions and the time between failures as covariates of the
system. The latter can be described by the Generalized Proportional Intensity Model (GPIM)
which can also incorporate in its’ intensity function on the effect of preventive and corrective
maintenance actions. The results of the GPIM model found by the maximum likelihood
approach make it possible to begin the frequency’s optimization of preventive maintenance
using a simulation algorithm on a fixed time scale. The calculation of the average cost by this
algorithm requires a large number of simulations (n>1000), which implies the use of software
such as the MATLAB programming language. The proposed periodicity must reflect the
minimum possible costs while ensuring the optimal balance between preventive and
corrective maintenance actions.
