Neural-Network Analysis of Irradiation Hardening in Low-Activation Steels

R. Kemp, G. A. Cottrell, H. K. D. H. Bhadeshia, G. R. Odette, T. Yamamoto and H. Kishimoto

Abstract

An artificial neural network has been used to model the irradiation hardening of low-activation ferritic/martensitic steels. The data used to create the model span a range of displacement damage of 0 - 90 dpa, within a temperature range of 273 - 973 K and contain 1800 points. The trained model has been able to capture the non-linear dependence of yield strength on the chemical composition and irradiation parameters. The ability of the model to generalise on unseen data has been tested and regions within the input domain that are sparsely populated have been identified. These are the regions where future experiments could be focused. It is shown that this method of analysis, because of its ability to capture complex relationships between the many variables, could help in the design of maximally informative experiments on materials in future irradiation test facilities. This will accelerate the acquisition of the key missing knowledge to assist the materials choices in a future fusion power plant.

Journal of Nuclear Materials 348 (2006) 311-328

Download PDF file of paper.

Related paper.

Related software.

Related data. The experimental data were originally assembled by G. R. Odette and T. Yamamoto, UCSB. Please contact yamataku@engineering.ucsb.edu for fuller information. Only one copy should be downloaded for personal use and the data should not be further disseminated.

Neural networks.








Superalloys Titanium Bainite Martensite Widmanstätten ferrite
Cast iron Welding Allotriomorphic ferrite Movies Slides
Neural Networks Creep Mechanicallly Alloyed Theses Retained Austenite

PT Group Home Materials Algorithms