Analysing the Structure of Poly-crystalline Materials by 2-Dimensional DLS-Spectra and Neural Nets

摘要

A method to analyse the structure of poly-crystalline material by 2-dimensional DLS-Spectra and Backpropagation neural nets is introduced. It will be shown that DLS-Spectra ''clean up'' X-Ray-pictures such that the Radon-transformation of the Kikuchi-diagrams leads to better results. Further it will be shown, how these cleaned up Kikuchi-diagrams enable an automatic classification of poly-crystalline materials by neural hyper-classification systems based on the Backpropagation-algorithm.