Energy Consumption and CO 2 Emissions Forecasting of Transport Sector Using Machine Learning
Energy Science & Engineering · 2026-07 · 14 (7) · 3508-3526
DOI: 10.1002/ese3.70545 ↗Kurz gefasst
Fachartikel in der Zeitschrift Energy Science & Engineering, online erschienen am 02.05.2026. Verfasst von Amir Hossein Akbari und 2 weiteren Autor:innen. Die Arbeit nennt 72 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The transport sector accounts for approximately one‐quarter of Iran's final energy consumption. The energy demand in this sector has the least variation, with petroleum products accounting for more than 85% of the demand. Furthermore, the accelerated growth of energy consumption and the sector's reliance on fossil fuels, which are the main cause of pollution, are of significant concern to policymakers. This study employs various machine learning techniques—including linear regression, linear‐logarithmic regression, lasso regression, ridge regression, and multi‐layer perceptron—to analyze 1967–2019 data using per capita GDP, population, vehicle fleet size, vehicle fleet average age, and fuel price as inputs. The optimized MLP model exhibited the highest predictive accuracy ( R ² = 0.996, low MSE/MAPE) and was subsequently employed to forecast the energy consumption and CO₂ emissions of Iran's road transport sector through 2030 under three policy scenarios: The continuation of the current process, without the entry of electric vehicles, represents the baseline scenario, designated as “a.” This is followed by two alternative scenarios that utilize the 2040 policy targets as boundary conditions for the 2019–2030 forecast: Scenario b models EV entry following Approach‑1 (reaching an annual entry of 100,000 EVs by 2040), and Scenario c follows Approach‑2 (assuming a complete phase‑out of gasoline vehicle sales by 2040). Although the scenario targets extend to 2040, their implications are evaluated within the medium‐term forecast horizon up to 2030. Scenario projects gasoline at 51.44% of the fuel portfolio by 2030. Scenarios b and c yield 51.23% and 48.89% respectively, with CO₂ reductions of 0.45% and 5.35% relative to the baseline. The limited reductions observed in the gradual adoption scenario reflect the slow turnover of the vehicle fleet in the short to medium term. These outcomes highlight the importance of accelerated electrification pathways for transport‐sector decarbonization.
Zitieren
APA
Akbari, A. H., Hashemi, S. M., & Farajollahi, M. (2026). Energy Consumption and CO 2 Emissions Forecasting of Transport Sector Using Machine Learning. Energy Science & Engineering, 14(7), 3508-3526. https://doi.org/10.1002/ese3.70545
Vancouver
Akbari AH, Hashemi SM, Farajollahi M. Energy Consumption and CO 2 Emissions Forecasting of Transport Sector Using Machine Learning. Energy Science & Engineering. 2026;14(7):3508-3526. doi:10.1002/ese3.70545
BibTeX
@article{Akbari2026,
author = {Akbari, Amir Hossein and Hashemi, Seyyed Mohammad and Farajollahi, Meisam},
title = {Energy Consumption and CO 2 Emissions Forecasting of Transport Sector Using Machine Learning},
journal = {Energy Science & Engineering},
year = {2026},
volume = {14},
number = {7},
pages = {3508--3526},
doi = {10.1002/ese3.70545}
}
EUVIMED