@article{ART003259668},
author={Hyun-Ju Kim and Dong-Hyun Kim and Kyong-Hyon Kim},
title={Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM},
journal={Journal of Internet of Things and Convergence},
issn={2466-0078},
year={2025},
volume={11},
number={5},
pages={10}
TY - JOUR
AU - Hyun-Ju Kim
AU - Dong-Hyun Kim
AU - Kyong-Hyon Kim
TI - Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM
JO - Journal of Internet of Things and Convergence
PY - 2025
VL - 11
IS - 5
PB - The Korea Internet of Things Society
SP - 10
EP -
SN - 2466-0078
AB - In response to the International Maritime Organization’s (IMO) mid-term greenhouse gas (GHG) strategy, all vessels of 5,000 gross tonnage or more will be required to collect fuel intensity (GFI) data from 2028, underscoring the growing importance of fuel efficiency enhancement and accurate carbon emission prediction. This study aims to address these regulatory demands by applying a Bi-LSTM (Bidirectional Long Short-Term Memory) model, which is well-suited for time-series analysis, to precisely predict CO₂ emissions from LNG dual-fuel (DF) engine vessels while ensuring model interpretability. 230,000 real voyage data points collected between February and September 2023 were processed, and about 10,000 valid normal navigation records (above 6.75 knots) were selected for model training. The target variable was defined as the hourly CO₂ emission rate based on LSFO and LNG fuel consumption. The Bi-LSTM model achieved superior predictive performance, recording an R² of 0.9761, RMSE of 0.2113, and MAPE of 4.60%. Furthermore, SHAP (SHapley Additive exPlanations) analysis identified vessel speed (SPEED_VG) and cargo vapor temperature (CARGO_VAPOR_TEMP_MEAN) as the most influential variables, quantitatively demonstrating that increases in these factors contribute to higher CO₂ emission predictions.
KW - LNG dual-fuel ships;CO₂ emission prediction;Bi-LSTM;SHAP analysis;Time-series data
DO -
UR -
ER -
Hyun-Ju Kim, Dong-Hyun Kim and Kyong-Hyon Kim. (2025). Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM. Journal of Internet of Things and Convergence, 11(5), 10.
Hyun-Ju Kim, Dong-Hyun Kim and Kyong-Hyon Kim. 2025, "Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM", Journal of Internet of Things and Convergence, vol.11, no.5 10.
Hyun-Ju Kim, Dong-Hyun Kim, Kyong-Hyon Kim "Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM" Journal of Internet of Things and Convergence 11.5 10 (2025) : 10.
Hyun-Ju Kim, Dong-Hyun Kim, Kyong-Hyon Kim. Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM. 2025; 11(5), 10.
Hyun-Ju Kim, Dong-Hyun Kim and Kyong-Hyon Kim. "Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM" Journal of Internet of Things and Convergence 11, no.5 (2025) : 10.
Hyun-Ju Kim; Dong-Hyun Kim; Kyong-Hyon Kim. Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM. Journal of Internet of Things and Convergence, 11(5), 10.
Hyun-Ju Kim; Dong-Hyun Kim; Kyong-Hyon Kim. Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM. Journal of Internet of Things and Convergence. 2025; 11(5) 10.
Hyun-Ju Kim, Dong-Hyun Kim, Kyong-Hyon Kim. Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM. 2025; 11(5), 10.
Hyun-Ju Kim, Dong-Hyun Kim and Kyong-Hyon Kim. "Predicting Carbon Emissions from LNG Dual-Fuel Vessels and Analyzing Feature Importance with Bi-LSTM" Journal of Internet of Things and Convergence 11, no.5 (2025) : 10.
