University of Twente Student Theses
Multimodal Neural Networks and Data Preprocessing for Ship Trajectory Prediction Using AIS, Radar, ENC and Weather Data
Sokolovas, Erikas (2023) Multimodal Neural Networks and Data Preprocessing for Ship Trajectory Prediction Using AIS, Radar, ENC and Weather Data.
Full text not available from this repository.
| Full Text Status: | Access to this publication is restricted |
| Embargo date: | 16 November 2028 |
| Abstract: | Increasing maritime vessel traffic traffic in busy waterways due to increased international seaborne trade is straining existing Vessel Traffic Service (VTS) systems - requiring new technical solutions to manage the increased traffic. Ship Trajectory Prediction (STP) is likely to be one such component used for safety or optimization. Most current STP models only utilise Automatic Identification System (AIS) messages transmitted by vessels. However, AIS is known as an important, but low quality, data source. In this work we attempt to improve upon current STP models by utilising radar, Electronic Navigation Chart (ENC) and weather as data source in addition to AIS by developing multi-modal RNN and Transformer deep learning models for use in the STP task. We investigate the potential cross-region extendability capabilities of multi-modal STP models. We find that with sufficient cleaning and engineering of AIS data there is no model prediction accuracy improvement from the additional data sources. Additionally, we find that Transformer neural networks perform slightly better at the STP task than RNN’s. We also observe some evidence to suggest that the multi-modal STP models might be capable of cross-region extendability. |
| Item Type: | Essay (Master) |
| Clients: | Tidalis BV, Apeldoorn, Netherlands |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science, 55 traffic technology, transport technology |
| Programme: | Computer Science MSc (60300) |
| Link to this item: | https://purl.utwente.nl/essays/97614 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page