Scholarlink

<p>The present work investigated ethanol–water mixture for internal combustion engines to attain maximum efficiency with the least emission to tackle the challenge presented, considering the current environmental and transportation regulations. Ethanol and water mixtures, due to their favorable features like high-octane values and improved anti-knock ability, are reported as effective substitutes for the traditional combustible fuels under high compression ratios. The eXtreme Gradient Boosting and Support Vector Regression prediction models have been applied to estimate emissions and the function of the engines. By the process of hyperparameter tuning, the models showed more accurate predictions for the key pollutants CO, NO<sub>x</sub>, and HC. Specifically, the XGBoost model created an R<sup>2</sup> value equal to 0.998 for the prediction of NOx, whereas the MAPE was 0.078%, and the SVR created an R<sup>2</sup> value equal to 0.966 with an MAPE of 0.171% for the HC emission. The comparison study revealed that XGBoost was better predictive in comparison to SVR. The study emphasized the significance of feature engineering and variability control to ensure robustness during the prediction. The present work provided insightful observations on the optimization of the environment on the engine and the fuel composition to reach a balance among the emissions, efficiency, and performance, and hence it supported the evolution towards greener and efficient technologies for the engine.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Exploring advanced predictive models for emission forecasting and performance optimization in ethanol–water fueled engines

  • 期刊论文

  • 发表日期：

    2025年12月15日

查看全文
  • Dandan Song,
  • Tao Yang

摘要

The present work investigated ethanol–water mixture for internal combustion engines to attain maximum efficiency with the least emission to tackle the challenge presented, considering the current environmental and transportation regulations. Ethanol and water mixtures, due to their favorable features like high-octane values and improved anti-knock ability, are reported as effective substitutes for the traditional combustible fuels under high compression ratios. The eXtreme Gradient Boosting and Support Vector Regression prediction models have been applied to estimate emissions and the function of the engines. By the process of hyperparameter tuning, the models showed more accurate predictions for the key pollutants CO, NOx, and HC. Specifically, the XGBoost model created an R2 value equal to 0.998 for the prediction of NOx, whereas the MAPE was 0.078%, and the SVR created an R2 value equal to 0.966 with an MAPE of 0.171% for the HC emission. The comparison study revealed that XGBoost was better predictive in comparison to SVR. The study emphasized the significance of feature engineering and variability control to ensure robustness during the prediction. The present work provided insightful observations on the optimization of the environment on the engine and the fuel composition to reach a balance among the emissions, efficiency, and performance, and hence it supported the evolution towards greener and efficient technologies for the engine.