Designing a Bayesian network for preventive maintenance from expert opinion in a rapid and reliable way
In this study, a Bayesian Network (BN) is considered to represent a nuclear plant mechanical system degradation. It describes a causal representation of the phenomena involved in the degradation process. Inference from such a BN needs a great number of marginal and conditional probabilities have to be provided. As, in the present context, information is based essentially on expert knowledge, this task becomes very complex and rapidly impossible. We present a solution which consists of considering the BN as a log-linear model on which simplification constraints are assumed. This approach results in a considerable decrease in the number of probabilities to be given by expertise. We give some rules to choose the most reliable probabilities. Acting in such a way, consistency of the required probabilities can be checked. Moreover, a feedback procedure is proposed to eliminate inconsistent probabilities. Finally, the retained probabilities to solve the system are expected to be the most reliable ones. This model is applied to a reactor coolant sub-component in EDF Nuclear plants in an illustrative purpose.