ENG
INA
89th Doctoral Promotion
Promovendus Widyadi Setiawan succeeded in his doctoral dissertation exam and became the 89th PhD of the Study Program of Doctoral Engineering Science (PSDIT), Faculty of Engineering, Udayana University. The dissertation titled Artificial Intelligence Ssystem for Anomalous Behavior Identification Based on Automatic Identification System Data received a very satisfactory predicate.
Dr. Widyadi Setiawan described that the increase in maritime activities raises an urgent need to detect anomalies in vessel behavior based on AIS (Automatic Identification System) data. This research aims to develop a framework to identify vessel trajectories with potential anomalous behavior, such as illegal transshipment, loitering, and AIS device on-off activities.
The framework starts with the AIS data preprocessing stage, which includes data reading, data cleaning, trajectory extraction, and trajectory cleaning. Next, anomaly identification uses a combined Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and Neural Network (NN). By utilizing parallel processing, the data volume was reduced from 230 million to 123 million coordinate points. Integrating AIS data with trajectory extraction techniques enables large-scale analysis by effectively linking AIS data and trajectory data to support the identification of potential anomalies.
The developed model successfully identified anomalies with this approach. The transshipment model identified 8,651 track pairs with criterion 1, 1,592 with criterion 2, and 19 with criterion 3. The loitering model found 24,042 tracks out of 156,355, while the on-off model detected 572,808 coordinates as anomalies out of 123,632,829. DBSCAN successfully separated outliers from the data optimally using an epsilon parameter of 0.95 and a minimum sample size of 6. Meanwhile, the NN model showed high performance in all three anomaly categories studied. Validation of the framework showed that transshipments remained at 19, reducing loitering trajectories from 24,042 to 8,984 and on-off anomalies from 572,808 to 568,989. The results show that the rule-based preprocessing method with DBSCAN and NN approach can provide a solution for anomaly identification in AIS data.
The promoter, co-promoters I and II, are Prof. Ir. Linawati, M.Eng.Sc., PhD., IPU, Prof. Dr. Ir. I Made Oka Widyantara, ST.,MT., IPU, ASEAN. Eng and Dr. Dewa Made Wiharta, ST., MT, respectively.
FACULTY OF ENGINEERING