Apports des réseaux bayésiens à la prévention du risque de piraterie à l'encontre des plateformes pétrolières

摘要

In recent years, pirate attacks against ships or oil fields have continued to multiply and worsen. For example, the attack against the Exxon Mobil platform in 2010 in the coast of Nigeria has resulted in the removal of nineteen crew members and the reduction of 45,000 barrels of daily oil production which resulted in a rise prices internationally.This example is a perfect illustration of the extent of damage on the safety of offshore oil infrastructure.As part of our research, we propose an approach to control and management of the risk of piracy based on the concept of Bayesian networks that enable knowledge representation and calculation of conditional probabilities.A temporal dimension was added by the use of Bayesian networks called "dynamic". These networks, based on Markov chains hidden or Kalman filters, are proving very effective in the field of risk analysis.The application of these networks on piracy was considered, the contributions and limitations will be evaluated as part of this thesis.