University of Twente Student Theses

Login

Viability of anomaly detection aid for vessel traffic service operators

Hendriksen, S.J.F. (2022) Viability of anomaly detection aid for vessel traffic service operators.

[img] PDF
1MB
Abstract:This work examines the path toward viability for maritime anomaly detection as an aid for vessel traffic service operators, building from the literature towards a proof of concept and evaluating hurdles in the process. Leveraging data from moving vessels the work discusses maritime anomaly grouping, detection strategy, transparency, and visualization as part of the proof of concept. Anomaly types defined in the framework supply the foundation for the detection strategy used in the proof of concept. The selected maritime anomaly detection strategy, leveraging self-organizing maps and Gaussian mixture model normalized probability density, performs well by consistently highlighting low-density areas. Separate transparency models trained on subsets of the features score individual anomaly types. These transparency models enable the corresponding visualization to provide what anomaly types deviate a vessel from normalcy. The proof of concept and process are evaluated in one-on-one expert sessions, resulting in three aspects to mature the maritime anomaly detection field towards real-world application. These aspects are: defining the behavior constituting a maritime anomaly, focusing on transparency as part of the strategy, and selecting the relevant anomaly types for vessel traffic service operators.
Item Type:Essay (Master)
Clients:
Saab Technologies B.V., Apeldoorn, Nederland
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science, 55 traffic technology, transport technology
Programme:Business Information Technology MSc (60025)
Link to this item:https://purl.utwente.nl/essays/93716
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page