This thesis presents the design, optimisation and testing of a novel in-field system for illicit drug identification and its implementation to a fast-parcel environment using Energy Dispersive X-ray Diffraction and multivariate analysis. A major threat in security in recent years has been the transportation of illicit drugs into the United Kingdom through a variety of routes, and most predominantly via the postal system. X-ray diffraction’s low false alarm rate and reduced need for manual inspections had been found to improve screening effectiveness and detection performance, making it a viable alternative for border control applications. A system was primarily designed to determine the optimal configuration of secondary collimators. Two out of six possible scattering cell designs, ‘V’ and ‘VV’, provided the highest counts per minute whilst keeping the peak resolution acceptable, and it was decided that both should be accommodated in the DILAX built. The prototype DILAX system was then manufactured and its mechanical components characterised and subsequently optimised. The inherent filtration present in the DILAX source resulted in significantly more beam-hardening compared to the design. Furthermore, the prototype featured an extrapolation from a single diffraction beam, used in the design, to an array of twenty diffraction beams and equivalent collimator-detector systems, which led to cross-contamination due to multiple scatter recorded. However, the potential of the system was determined by evaluating its diffraction capabilities and associated prediction power under a fast-parcel setting using various material libraries. The diffraction profiles of drugs and typical cutting agents were recorded to cover as many illicit substances and adulterants as possible for a conclusive system, and the possibility of combining diffraction information with transmission images to improve performance has been demonstrated. Multivariate analysis was performed on a library of seventy-five simulated parcels, with the DILAX system scoring relatively high sensitivity and specificity at 83.02% and 77.27% respectively with a total accuracy of 81.33%.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:668457 |
| Date | January 2015 |
| Creators | Drakos, I. |
| Contributors | Speller, R. |
| Publisher | University College London (University of London) |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://discovery.ucl.ac.uk/1468928/ |
Page generated in 0.0019 seconds