This thesis supports the ARCOS project, which was created to safeguard and monitor the Arctic region that has been affected by climate problems in recent years but, at the same time, represents a destination for new economic opportunities. The algorithm discussed in this thesis aims at monitoring ships transiting the Arctic region, identifying the most frequently sailed routes and any outlier routes. The geographical coordinates of the ships are extrapolated, along with other information, from the Automatic Identification System (AIS) data transmitted by the ships themselves. The algorithm developed in Python has several objectives: (i) to examine a cluster of ship routes (trajectories) and split them into summer and winter seasons; and (ii) to subdivide the trajectories according to their similarity, using a trajectory clustering algorithm aimed at dividing them into subclusters. Once the subclusters are created, an algorithm is proposed to extrapolate the “realistic average trajectory” of each subcluster. To extrapolate such average route, first a ”normalization” algorithm is introduced, aimed at transforming the trajectories into a form that allows for correct and efficient analysis. Then, it is introduced an algorithm that computes the “unrealistic average trajectory”, aimed at computing the trajectory that is most structurally similar to the “realistic average trajectory” to be computed. An algorithm is then explained to score all docking ports of ships belonging to a subcluster. Finally, the results computed by the various algorithms are displayed on a geographic map in the format produced by the Folium library and in the QGIS application through GeoJSON format. Through the results shown on a geographical map, the subcluster division made by the algorithm is examined and the most navigated and outlier routes are highlighted. For each subcluster, the average route computed by the algorithm and all docking ports are also examined. The “normalization” algorithm and the “unrealistic average trajectory” computation algorithm are then discussed. Finally, graphs are provided showing the characteristics of the ships belonging to the clusters and the execution times of the algorithm started in the multithreading modes enabled and disabled.
Questa tesi supporta il progetto ARCOS, nato per salvaguardare e monitorare la regione Artica che negli ultimi anni è afflitta da problemi climatici ma, contestualmente, rappresenta una meta per nuove opportunità economiche. L’algoritmo qui trattato mira a monitorare le navi in transito nella regione Artica, identificando le rotte più frequentemente navigate ed eventuali rotte anomale. Le coordinate geografiche delle navi vengono estrapolate, insieme ad altre informazioni, dai dati del sistema di identificazione automatica (AIS) trasmessi dalle navi medesime. L’algoritmo sviluppato in Python ha diversi obiettivi: (i) esaminare un cluster di rotte navali (traiettorie) e scinderle in stagione estiva e invernale; e (ii) suddividere le traiettorie in base alla loro similarità, mediante un algoritmo di clustering di traiettorie volto a dividerle in sotto-cluster. Una volta creati i sotto-cluster, si propone un algoritmo che permette di estrapolare la “traiettoria media realistica” di ogni sotto-cluster. Per estrapolare tale rotta media, si introduce prima un algoritmo di “normalizzazione”, volto a trasformare le traiettorie in una forma che consenta un’analisi corretta ed efficiente, e successivamente un algoritmo che calcola la “traiettoria media non realistica”, volto a computare la traiettoria più simile strutturalmente alla “traiettoria media realistica” da calcolarsi. Si spiega poi un algoritmo che permette di segnare tutti i porti di attracco delle navi appartenenti a un sotto-cluster. Infine, i risultati calcolati dai vari algoritmi vengono mostrati su mappa geografica nel formato prodotto dalla libreria Folium e nell’applicazione QGIS in formato GeoJSON. Mediante i risultati riportati su mappa geografica, si esamina la divisione in sotto-cluster effettuata dall’algoritmo e si evidenziano le rotte più navigate e quelle anomale. Per ogni sotto-cluster si esamina altresì la rotta media computata dall’algoritmo e tutti i porti di attracco. Si discute poi, tra l’altro, l'algoritmo di ”normalizzazione” e quello di calcolo della “traiettoria media non realistica”. Infine, si riportano grafici con le caratteristiche delle navi appartenenti ai cluster e i tempi di esecuzione dell’algoritmo avviato nelle modalità multithreading attivata e disattivata.
A methodology to extract naval routes from AIS data: empirical analisys of maritime traffic in the Arctic region
Secondari, Simone
2022/2023
Abstract
This thesis supports the ARCOS project, which was created to safeguard and monitor the Arctic region that has been affected by climate problems in recent years but, at the same time, represents a destination for new economic opportunities. The algorithm discussed in this thesis aims at monitoring ships transiting the Arctic region, identifying the most frequently sailed routes and any outlier routes. The geographical coordinates of the ships are extrapolated, along with other information, from the Automatic Identification System (AIS) data transmitted by the ships themselves. The algorithm developed in Python has several objectives: (i) to examine a cluster of ship routes (trajectories) and split them into summer and winter seasons; and (ii) to subdivide the trajectories according to their similarity, using a trajectory clustering algorithm aimed at dividing them into subclusters. Once the subclusters are created, an algorithm is proposed to extrapolate the “realistic average trajectory” of each subcluster. To extrapolate such average route, first a ”normalization” algorithm is introduced, aimed at transforming the trajectories into a form that allows for correct and efficient analysis. Then, it is introduced an algorithm that computes the “unrealistic average trajectory”, aimed at computing the trajectory that is most structurally similar to the “realistic average trajectory” to be computed. An algorithm is then explained to score all docking ports of ships belonging to a subcluster. Finally, the results computed by the various algorithms are displayed on a geographic map in the format produced by the Folium library and in the QGIS application through GeoJSON format. Through the results shown on a geographical map, the subcluster division made by the algorithm is examined and the most navigated and outlier routes are highlighted. For each subcluster, the average route computed by the algorithm and all docking ports are also examined. The “normalization” algorithm and the “unrealistic average trajectory” computation algorithm are then discussed. Finally, graphs are provided showing the characteristics of the ships belonging to the clusters and the execution times of the algorithm started in the multithreading modes enabled and disabled.File | Dimensione | Formato | |
---|---|---|---|
TESI.pdf
accessibile in internet per tutti
Descrizione: TESI DI LAUREA MAGISTRALE IN COMPUTER SCIENCE AND ENGINEERING
Dimensione
4.16 MB
Formato
Adobe PDF
|
4.16 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/196513