Перейти к содержимому
Intelligent Management and Artificial Intelligence: Trends, Challenges, and Opportunities, Vol.2

Proceedings on 28th European Conference on Artificial Intelligence ECAI 2025 – InMan Workshop

Intelligent Management and Artificial Intelligence: Trends, Challenges, and Opportunities, Vol.2
OpenAlex

Данные относятся к этой статье или главе.

Запись найдена в базе

Цитирования
0
Библиографические ссылки
18

Данные получены:

Показаны последние доступные данные. Обновление ожидается или не удалось.

Открыть источник (новая вкладка)
OpenCitations

Данные относятся к этой статье или главе.

Цитирования
0

Данные получены:

Показаны последние доступные данные. Обновление ожидается или не удалось.

Открыть источник (новая вкладка)
Google Scholar

LARGE DATABASE OF AIS VESSEL TRAFFIC FOR SMART DECISION-MAKING BASED ON AI IN SHIPPING COMPANIES

DOI: 10.18276/978-83-8419-053-1-29
Цитировать

Полный текст доступен

Абстракт

Purpose: This study aims to assess the effectiveness of Principal Component Analysis (PCA) in identifying anomalies and improving data quality in AIS (Automatic Identification System) datasets used for maritime monitoring and decision-making. Need for the study: AIS plays a crucial role in intelligent vessel traffic management across the shipping industry. However, despite its widespread adoption, the technology still suffers from issues of data consistency, accuracy, and integrity. These limitations reduce its effectiveness in supporting AI-based decision-making systems. Addressing these shortcomings is essential for improving operational efficiency, safety, and sustainability in maritime transport. Methodology: The research utilizes PCA on a large-scale AIS dataset containing over 800 million records from the Baltic and North Seas. The data underwent filtering and cleaning before analysis to detect patterns, identify anomalies, and evaluate the consistency of variables such as latitude, longitude, draught, and speed. Additionally, t-SNE (t-distributed Stochastic Neighbor Embedding) and UMAP (Uniform Manifold Approximation and Projection) were explored as complementary nonlinear dimensionality reduction techniques. Findings: PCA proved effective in detecting both extreme and subtle anomalies in the AIS data, including implausible draught values, unusual routing behaviors, and spatial inconsistencies. The analysis demonstrated that even mature systems like AIS can yield unreliable data, emphasizing the necessity for rigorous quality control. Furthermore, the study found that nonlinear methods like t-SNE and UMAP hold promise for enhancing dynamic monitoring. Practical Implications: The integration of PCA-based analytics into maritime monitoring systems enhances early anomaly detection, reduces the risks of decision-making based on faulty data, and improves situational awareness. The study advocates for stronger AIS data governance standards and lays the foundation for future use of real-time anomaly detection and AI-driven decision support in maritime operations.
Просмотр PDF

Открыть PDF