Apr 15 – 16, 2019
University of Surrey
Europe/London timezone

Enhancing source localisation for threat detection

Apr 15, 2019, 5:30 PM
University of Surrey

University of Surrey

Guildford, UK
Pilot Project Poster Detectors & Systems Session 4: Poster session and drinks reception


Antonin Vacheret (Imperial College London)


Highly sensitive neutron detectors with directional information could improve the timeliness of detection and localisation of special nuclear material (SNM) at national ports of entry and for deriving operational and protection quantities in the field of neutron dosimetry.
In this context, two neutron source localisation algorithms, for use with a novel plastic segmented scintillator detector, nFacet 3D, were developed and evaluated as part of this study. Experimental data from Cf, AmBe and AmLi sources at different angles to the nFacet 3D detector were collected at the National Physical Laboratory (UK). Detection and localisation algorithms based on multilayer perceptron (MLP) and convolutional neural network (CNN) architectures were modelled and tested on data simulated from experimental records.
Results of this study includes prediction of the presence of a neutron source, estimation of a direction towards its location, and identification of the source from a set of given possibilities, supporting the feasibility of using machine learning methods in autonomous decision-making for threat detection.

Primary authors

Antonin Vacheret (Imperial College London) Shameena Bonomally (Imperial College (GB)) Dr Sakari Ihantola (STUK)

Presentation materials