Sample positioning in neutron diffraction experiments using a multi-material fiducial marker

Abstract

An alternative sample positioning method is reported for use in conjunction with sample positioning and experiment planning software systems deployed on some neutron diffraction strain scanners. In this approach, the spherical fiducial markers and location trackers used with optical metrology hardware are replaced with a specifically designed multi-material fiducial marker that requires one diffraction measurement. In a blind setting, the marker position can be determined within an accuracy of ±164 µm with respect to the instrument gauge volume. The scheme is based on a pre-determined relationship that links the diffracted peak intensity to the absolute positioning of the fiducial marker with respect to the instrument gauge volume. Two methods for establishing the linking relationship are presented, respectively based on fitting multi-dimensional quadratic functions and a cross-correlation artificial neural network.

Derivation of NARX models by expanding activation functions in neural networks

“…Chen et al [3] and Levin and Narendra [4] demonstrated that nonlinear systems can be identified using neural networks. Furthermore, free open-source libraries such as the Fast Artificial Neural Network Library (FANN) [5] for network learning have already enabled researchers in various fields to use neural networks [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. In fact, neural networks have recently been used for the identification of a wide range of nonlinear systems, including biological systems [23][24][25][26][27][28][29][30][31][32][33][34][35][36].…”