Sélection de la langue

Gouvernement du Canada / Government of Canada

Recherche

A unified data platform for maritime analytics and predictive modeling in Arctic operations

Par Conseil national de recherches du Canada

Lienhttps://www.poac.com/Papers/2025/pdf/POAC25_paper_28.pdf
AuteurRechercher : 1; Rechercher : 1Identifiant ORCID : https://orcid.org/0000-0002-3371-1082; Rechercher : 1; Rechercher : ; Rechercher : 1
Affiliation
  1. Conseil national de recherches Canada. Génie océanique, côtier et fluvial
FormatTexte, Article
Conférence28th International Conference on Port and Ocean Engineering under Arctic Conditions, POAC, July 13-17, 2025, St. Johns, Newfoundland and Labrador, Canada
Sujetmaritime analytics; machine learning; emissions tracking; data visualization; operational efficiency
Résumé
As the shipping industry embraces digitalization, marine operations generate vast and diverse data streams, offering significant opportunities for advanced analytics. This work presents an advanced software platform powered by a high-performance Online Analytical Processing (OLAP) database, centralizing data from multiple Canadian Coast Guard (CCG) vessels, including the Larsen, Laurier, Cygnus, Tully, and Tanu. The system integrates operational data (e.g., fuel consumption, navigation) and environmental datasets (e.g., ERA5 weather, ocean currents, ice charts) while harmonizing sensor nomenclature for consistency. For example, vessel-specific labels such as "ENG_RPM" and "ENGINE_SPEED" are standardized to a unified "Engine RPM" field using a middle vocabulary layer between database and, enabling consistent cross-vessel comparisons. The platform's visualization tool enables quality assurance (QA), exploratory data analysis (EDA), and comparative analysis across vessels using scatter plots, trajectory maps, and fuel consumption summaries. Its emissions module provides detailed hourly and trip-based emissions calculations, supporting regulatory compliance and environmental assessments. Benchmarking features include univariate and bivariate plots, time series boxplots, and heatmaps for cross-vessel comparisons under varying conditions. A key innovation is its accessible machine learning (ML) tool, allowing non-expert users to train predictive models on custom time intervals for forecasting metrics like fuel consumption, speed, and emissions. A dedicated ML Model Explainer dashboard enhances interpretability with regression statistics, feature importance rankings, and dependency charts. Additionally, the platform supports strategic planning through historical route mapping and predictive analytics, enabling users to test models under Arctic conditions. By integrating visualization, benchmarking, and ML-driven forecasting, this platform enhances maritime decision-making, optimizing operational efficiency and environmental sustainability across challenging marine environments.
Date de publication
Maison d’éditionLulea University of Technology
POAC Secretary General
Dans
Langueanglais
Publications évaluées par des pairsOui
Exporter la noticeExporter en format RIS
Signaler une correctionSignaler une correction (s'ouvre dans un nouvel onglet)
Identificateur de l’enregistrement861b6326-89ab-4ee3-92e7-f043e233b58f
Enregistrement créé2025-09-11
Enregistrement modifié2025-09-24
Date de modification :