Study on CO2 Emission Assessment of Heavy-Duty and Ultra-Heavy-Duty Vehicles Using Machine Learning

• Published in: International journal of automotive technology Vol. 25; no. 3; pp. 651 - 661
• Main Authors: Moon, Seokho, Lee, Jinhee, Kim, Hyung Jun, Kim, Jung Hwan, Park, Suhan
• Format: Journal Article
• Language: English
• Published: Seoul The Korean Society of Automotive Engineers 01.06.2024; Springer Nature B.V; 한국자동차공학회
• Subjects: Accuracy; Automotive Engineering; Carbon dioxide; Commercial vehicles; Comparative analysis; Emissions; Energy consumption; Engine and Emissions; Engine control; Engineering; Fuel consumption; Fuels and Lubricants; Heavy vehicles; Machine learning; Pollution monitoring; Regulations; Reliability; Road conditions; Trucks; 자동차공학; Machine learning; Heavy-duty vehicle; On-board monitoring; CO; emissions; Artificial intelligence prediction; Real driving emission; Portable emission measurement system; On-board diagnostics
• ISSN: 1229-9138; 1976-3832
• DOI: 10.1007/s12239-024-00051-5

EU is actively moving towards the implementation of Euro-7 regulations, thus placing a strong emphasis on real-road emissions. Euro-7 introduced OBM (on-board monitoring), which is an enhancement of regulations that closely replicates real-world road conditions. Furthermore, there is a need to devise an effective application strategy for utilizing the driving monitoring data prior to the enforcement of OBM. This study addresses these challenges by conducting RDE (real-driving emission) tests on both 3.5-ton and 25-ton commercial vehicles to gather CO 2 emissions and engine control unit data accessible through an OBD (on-board diagnostics) port. To process the RDE data, an appropriate machine learning model, XGBoost, was selected and trained. The outcome of our CO 2 emission prediction for the two vehicles demonstrated that employing monitoring data yielded reliable estimates of actual road CO 2 emissions. Finally, a comparative analysis was conducted between the proposed monitoring approach and the fuel-based CO 2 monitoring method using the emission factor from EMEP/EEA air pollutant emission inventory guidebook 2019 utilizing fuel consumption data achieved through the OBFCM (on-board fuel and energy consumption monitoring) rule. Our method, which is based on predictive CO 2 emissions monitoring, exhibited significantly greater accuracy. This outcome underscores the necessity to adopt the proposed approach.