Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Nondestructive Characterization of Reactor Pressure Vessel Materials Using Neural Networks

: Fiedler, U.; Schurig, C.; Böhmert, J.

Saarton, L.A.; Zeedijk, H.B. ; Federation of European Materials Societies; Netherlands Society for Materials Science:
Materials, functionality & design. Proceedings of the 5th European Conference on Advanced Materials and Processes and Applications. Vol. 4: Characterization and production/design : EUROMAT 97, Maastricht, NL, 21 - 23 April 1997
Zwijndrecht: Netherlands Society for Materials Science, 1997
ISBN: 90-803513-4-2
European Conference on Advanced Materials and Processes and Applications (EUROMAT) <5, 1997, Maastricht>
Fraunhofer IZFP ()
neural net; neuronales Netzwerk; nondestructive testing; pressure vessel; zerstörungsfreie Prüfung

Monitoring of deterioration of the mechanical properties of the reactor pressure vessel material during service and even better during operation using nondestructive methods could be an effective tool for safety assessment of aged reactors. For a Russian, low alloyed Cr-Mo-V-steel (15Xh2MFA) comprehensive results of destructive measurements for the determination of mechanical properties were collected (9 parameters characterizing strength, hardness, toughness). At the same specimens nondestructive micromagnetic measurements (13 different testing values derived from Barkhausen noise) were carried out. Both testing series were correlated with neural network technique to evaluate the nondestructive characterization of the mechanical properties. The neural net of backpropagation type was capable to successfully compute the mechanical properties.