Gaseous fuels are gaining increasing attention for power generation in internal combustion engines due to their cleaner combustion and potential for decarbonization. This study investigates the performance, combustion, and emission characteristics of a dual-fuel diesel engine operated with hydrogen, biogas, and ammonia as primary fuels, with diesel serving as the pilot fuel. At full engine load, the brake thermal efficiencies for hydrogen, biogas, and ammonia were 25.11 %, 19.25 %, and 17.33 %, respectively, compared to 23.11 % in diesel-only mode. The corresponding liquid fuel replacement ratios reached 79 %, 76.1 %, and 69 %, demonstrating significant potential for fossil fuel reduction. Combustion analysis revealed longer ignition delays for biogas (46.25 %) and ammonia (56.25 %) compared to hydrogen. Emission profiles indicated that hydrogen produced the lowest CO and HC emissions but exhibited higher NOx levels than the other fuels. In parallel, supervised machine learning models—Linear Regression, Extreme Gradient Boosting, and Gradient Boosted Regression Trees (GBRT)—were developed to predict brake thermal efficiency, liquid fuel substitution, peak cylinder pressure, and emissions (CO, HC, NOx). Tree-based ensemble models outperformed the linear baseline, effectively capturing the nonlinear influence of engine load and fuel lower heating value on engine responses. Gradient Boosted Regression Tree achieved the highest accuracy for brake thermal efficiencies (R2 = 0.9933) and CO (R2 = 0.9965), while Extreme Gradient Boosting was most accurate for peak cylinder pressure (R2 = 0.9898). SHAP-based explainable Artificial Intelligence analysis identified engine load as the dominant factor governing performance and emissions. Overall, the combined experimental–Machine Learning framework establishes hydrogen as a highly promising dual-fuel candidate and demonstrates Gradient Boosted Regression Trees’s strong capability for predictive optimization of dual-fuel combustion systems.
扫码关注我们
求助内容:
应助结果提醒方式:
