Source Localization via Neutron Imaging and Machine Learning

Neutron imaging systems are important diagnostic tools for characterizing the physics of inertial confinement fusion reactions at the National Ignition Facility (NIF). In particular, neutron images give diagnostic information on the size, symmetry, and shape of the fusion hot spot and surrounding cold fuel. Images are formed via collection of neutron flux from the source using a system of aperture arrays and scintillator-based detectors. Currently reconstruction of fusion source geometry from collected neutron images is accomplished by solving a Maximum Likelihood Expectation (MLE) problem via Expectation Maximization (EM). To accurately reconstruct the source geometry, the source location must be known. However, the knowledge of a priori source locations is limited by machine precision and experimental controls.

 

To overcome these physical limitations, the source location must be determined computationally from neutron image data. In this talk we show how to use neural networks to predict source locations in inertial confinement fusion reactions.  We provide experimental demonstrations of our methods on both non-noisy and noisy data. We also discuss the effects of training data sizes and the importance of choosing training data that covers the entire range of possible source locations.