Model reduction via principal component truncation for the optimal design of atmospheric monitoring networks
Résumé
IRSN is planning to renovate the French nuclear monitoring network. This renovated network should be able to forecast accurately the accidental plume by using only measures of this network. In this presentation, a numerically efficient methodology for the optimal design of monitoring networks is proposed. In this method, a large database of dispersion accidents over one year of meteorology and from 20 French nuclear sites is built and a cost function measures the ability of a potential network to provide measurements in order to reconstruct any accidental plume from the database. We introduce methods based on principal component analysis to optimally reduce this database and consequently decrease significantly CPU time. Then, the reduced optimisation method is applied to suggest an optimal strategy for the sequential deployment of the network. Finally, we propose the set-up of networks which take into account foreign potential radiological sources in Europe and French density population.