المفاهيم الأساسية
A data-driven framework for modeling, optimizing, and identifying vessel paths to improve energy efficiency in short-sea shipping.
الملخص
The paper presents a comprehensive framework for improving energy efficiency in short-sea shipping operations. Key highlights:
- Modeling of Energy Efficiency:
- Developed a data-driven model to estimate voyage energy efficiency, incorporating spatiotemporal aggregation of operational and environmental data.
- Introduced an efficiency score that considers both fuel consumption and voyage duration.
- Employed explainable AI (XAI) techniques to gain clear insights into the factors influencing energy efficiency.
- Voyage Optimization:
- Implemented four time-series analysis models (LSTM, kNN, 1NN-DTW, HMM) to optimize vessel speed profiles and improve energy efficiency.
- Evaluated the performance of these models across different data clusters to leverage insights from voyages with varying efficiency.
- Demonstrated the practical effectiveness of the approach for fixed-route vessels in short-sea shipping.
- Path Identification:
- Proposed a clustering-based framework to identify and label vessel paths, requiring only position information.
- Developed a robust and interpretable similarity measure to reduce the influence of noise and outliers.
- Provided a customizable parameter to determine the number of path clusters, enhancing flexibility.
- Analyzed patterns within specific segments of the vessel's path.
The framework integrates data-driven modeling, optimization, and path identification to deliver a comprehensive solution for improving energy efficiency in short-sea shipping operations.
الإحصائيات
The vessel's onboard data includes latitude, longitude, speed over ground, heading, pitch, roll, wind speed, and wind direction.
External weather data includes wind speed, wind direction, wave height, wave direction, current speed, and current direction.
اقتباسات
"The main outcomes of this paper can be summarized as follow:
Modeling of energy efficiency: Develop a data-driven model for voyage energy efficiency, including:
A spatiotemporal aggregation of operation and navigation data from onboard and external sources to capture the impact of both spatial and temporal factors on voyage energy efficiency.
Introduce an efficiency score that considers both total fuel consumption and voyage duration to measure the voyage energy efficiency."
"The proposed clustering approach of vessel paths requires only position information, specifically longitude and latitude.
The clustering approach has a proven added value for clustering challenging unseen or unknown paths.
The approach is robust and interpretable by applying a similarity measure that reduces the influence of noise or outliers and offers a clear interpretation of path clustering."