Machine learning prediction and evolutionary multi-objective optimization on performance and emissions of a syngas-diesel dual-fuel engine
International Journal of Hydrogen Energy, vol.162, 2025 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Volume: 162
- Publication Date: 2025
- Doi Number: 10.1016/j.ijhydene.2025.150781
- Journal Name: International Journal of Hydrogen Energy
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, PASCAL, Artic & Antarctic Regions, Chemical Abstracts Core, Chimica, Communication Abstracts, Compendex, Environment Index, INSPEC
- Keywords: Bayesian optimization, Dual-fuel engine, Engine performance, Machine learning, Multi-objective optimization, Syngas
- Erciyes University Affiliated: Yes
Abstract
This study introduces a machine learning-based framework to optimize performance and emissions in syngas–diesel dual-fuel compression-ignition engines. Experimental data from a single-cylinder engine testbed inform four predictive models— Artificial Neural Networks (ANN), Random Forest (RF), Support Vector Regression (SVR), and Extreme Gradient Boosting (XGBoost)—to estimate indicated thermal efficiency (ITE), CO, and NOx. Input variables include indicated mean effective pressure, fuel injection timing, hydrogen volumetric content, global excess air ratio, and syngas energy share. The ANN model demonstrates the highest predictive accuracy. Integrating this model with the Non-dominated Sorting Genetic Algorithm-III identifies optimal engine conditions, achieving 41.8 % ITE, 1970.5 ppm CO, and 746.5 ppm NOx emissions. A user-friendly graphical interface is developed for real-time predictions and decision-making. The proposed approach significantly advances clean engine technology by optimizing hydrogen-enriched syngas combustion, aligning closely with global sustainable energy objectives.