Bai, F.J.J.S.Varuvel, E.G.2024-05-192024-05-1920229786250008430https://hdl.handle.net/20.500.12713/4171BAU;et al.;INOGEN;Republic of Turkey, Ministry of Energy and Natural Resources;TENMARK;Turkish Airlines23rd World Hydrogen Energy Conference: Bridging Continents by H2, WHEC 2022 -- 26 June 2022 through 30 June 2022 -- -- 186176The transportation sector is a major emitter of carbon dioxide emissions. It is a known fact that carbon dioxide is the cause of global warming which has resulted in extreme weather conditions as well as climate change. In this study a combination of different methods of expediting the CO2 emission from a single cylinder common rail direct injection (CRDI) engine has been explored. The methods include use of low carbon content biofuels (lemon peel oil (LPO) and camphor oil (CMO), inducing hydrogen in the intake manifold and zeolite based after-treatment system. The emissions were found to reduce even further and at full load condition the lowest CO2 (39.7% reduction) and smoke (49% reduction) emissions were observed with LPO blend and hydrogen induction. The NO emission with hydrogen induction increases for both the blends, however, it was seen that the zeolite-based treatment system was effective in reducing the emission as well. As compared to baseline diesel, the maximum reduction in NO emission was 23% at full load with LPO blend, hydrogen induction and after-treatment system. Unfortunately, a lot of experimental research in the engineering field takes a long time and costs a lot of money. The number of experimental trials can be minimized by applying predictions using the available experimental data. Machine learning approaches can aid in the development of rapid and reliable data-based models that can supplement a traditional physical model. The goal of this work is to investigate if a data-based model can help to predict engine emission characteristics including CO, NOX, Smoke, BTE, and HC more accurately. In machine learning, ensemble methods are strategies for creating numerous models and then combining them to obtain better results. In most cases, ensemble approaches provide more accurate results than a single model. In the present work, ensemble learning algorithms like XGBoost, LightGBM, CatBoost, Random Forest (RF) are used for the prediction of emissions. The brake power and BSEC are the input parameters to these algorithms from which CO, NOX, Smoke, BTE, and HC are predicted under various combination of additives. The CatBoost model has produced high accuracy predictions which was followed by XGBoost, RF and LightGBM models. The predicted and actual values are compared each other and the performance of the algorithms were analysed using the evaluation metrics like R-Square(R2), Mean Absolute Error (MAE), Mean Square Error (MSE) and Root Mean Square Error (RMSE). © 2022 Proceedings of WHEC 2022 - 23rd World Hydrogen Energy Conference: Bridging Continents by H2. All rights reserved.eninfo:eu-repo/semantics/closedAccessAfter TreatmentEnsemble LearningHydrogenLow Carbon BiofuelMachine LearningConference Object133213342-s2.0-85147192562N/A