Bayesian approach and application to operation safety

The management of industrial risks requires the development of statistical and probabilistic analyses which use all the available convenient information in order to compensate the insufficient experience feedback in a domain where accidents and incidents remain too scarce to perform a classical statistical frequency analysis. The Bayesian decision approach is well adapted to this problem because it integrates both the expertise and the experience feedback. The domain of knowledge is widen, the forecasting study becomes possible and the decisions-remedial actions are strengthen thanks to risk-cost-benefit optimization analyzes. This book presents the bases of the Bayesian approach and its concrete applications in various industrial domains. After a mathematical presentation of the industrial operation safety concepts and of the Bayesian approach principles, this book treats of some of the problems that can be solved thanks to this approach: softwares reliability, controls linked with the equipments warranty, dynamical updating of databases, expertise modeling and weighting, Bayesian optimization in the domains of maintenance, quality control, tests and design of new equipments. A synthesis of the mathematical formulae used in this approach is given in conclusion. (J.S.)