No description available.

Cette thèse propose une méthodologie pour la conception de systèmes de diagnostic fiables permettant d'appréhender les systèmes dynamiques complexes et spatialement distribués. Les résultats proposés s'appuient d'une part sur des techniques d'analyse diagnostic formelle ou à base de consistance, qui permettent de garantir la justesse de l'analyse diagnostic, et d'autre part, sur le paradigme multi-agents. Les algorithmes proposés permettent de déduire tous les défauts possibles pour un comportement observé en les classant suivant différents critères de vraisemblance. Notre contribution a consisté à proposer une méthode de diagnostic qui tire partie des deux approches DX (communauté d'Intelligence Artificielle) et FDI (communauté Automatique) en distinguant la phase de détection, qui peut se faire à base de techniques variées et parfois très sophistiquées (observateur d'état, relation de parité, traitement de signal,...), de la phase de localisation ou d'analyse diagnostic, qui doit permettre de garantir ce qui peut l'être et d'analyser toutes les informations disponibles pour en déduire le diagnostic le plus juste et complet possible. De plus, nous avons montré qu'il était possible d'appréhender des incertitudes de décision en transposant la logique de l'analyse diagnostic en logique floue. Le travail présenté dans ce mémoire a été développé dans le cadre du projet européen MAGIC (Multi-Agents-Based Diagnostic Data Acquisition and Management in Complex systems). Les résultats sont en cours de transfert vers différentes industries grâce aux partenaires industriels du projet : les entreprises SATE (System Advanced Technologies Engineering) et SMS-DEMAG.

<p>Neutron-carbon interactions have been studied with a focus on charged-particle production of relevance to radiation protection and medical applications, such as cancer therapy. The measurements have been performed using the particle-detection setup, MEDLEY, and the 96 MeV neutron beam at the The Svedberg Laboratory in Uppsala.</p><p>Double-differential cross sections of inclusive charged-particle production are compared with recent calculations from models based on the GNASH code including direct, preequilibrium and compound processes. For protons, the shapes of the cross-section spectra are reasonably well described by the calculations. For the other particles- <i>d</i>, <i>t</i>, ³He and α- there are important discrepancies, in particular for ³He-ions and α-particles, concerning both shape and magnitude of the spectra.</p><p>Using the new cross sections, partial as well as total KERMA coefficients have been determined. The coefficients have also been compared to previous experimental results and model calculations. The <i>p</i>, <i>d</i> and <i>t</i> KERMA coefficients are in good agreement with those from a previous measurement. For the helium isotopes, there are no previous measurements at this energy. The KERMA coefficients are considerably higher (by up to 30%) than those predicted by the calculations.</p><p>The KERMA results indicate that protons and α -particles are the main contributors to the dose. A 6x6x6 cm³ carbon phantom, exposed to a broad and a pencil-like beam, is used for the computation of the absorbed doses deposited by these two particles in spheres of 1 μm in diameter, located at various positions in the phantom. The maximum doses are deposited at ~3 cm from the surface of neutron impact for protons and within 1 cm for α-particles. For the pencil beam, deposited doses are spread over regions of ~1.5 cm and ~300 μm transverse to the beam for protons and α-particles, respectively. The results are consistent with previous integral measurements at lower energies.</p>

Neutron-carbon interactions have been studied with a focus on charged-particle production of relevance to radiation protection and medical applications, such as cancer therapy. The measurements have been performed using the particle-detection setup, MEDLEY, and the 96 MeV neutron beam at the The Svedberg Laboratory in Uppsala. Double-differential cross sections of inclusive charged-particle production are compared with recent calculations from models based on the GNASH code including direct, preequilibrium and compound processes. For protons, the shapes of the cross-section spectra are reasonably well described by the calculations. For the other particles- d, t, 3He and α- there are important discrepancies, in particular for 3He-ions and α-particles, concerning both shape and magnitude of the spectra. Using the new cross sections, partial as well as total KERMA coefficients have been determined. The coefficients have also been compared to previous experimental results and model calculations. The p, d and t KERMA coefficients are in good agreement with those from a previous measurement. For the helium isotopes, there are no previous measurements at this energy. The KERMA coefficients are considerably higher (by up to 30%) than those predicted by the calculations. The KERMA results indicate that protons and α -particles are the main contributors to the dose. A 6x6x6 cm3 carbon phantom, exposed to a broad and a pencil-like beam, is used for the computation of the absorbed doses deposited by these two particles in spheres of 1 μm in diameter, located at various positions in the phantom. The maximum doses are deposited at ~3 cm from the surface of neutron impact for protons and within 1 cm for α-particles. For the pencil beam, deposited doses are spread over regions of ~1.5 cm and ~300 μm transverse to the beam for protons and α-particles, respectively. The results are consistent with previous integral measurements at lower energies.