Ammonia is a highly promising zero-carbon fuel for engines. However, it exhibits high ignition energy, slow flame propagation, and severe pollutant emissions, so it is usually burned in combination with highly reactive fuels such as hydrogen. An accurate understanding and modeling of ammonia–hydrogen combustion is of fundamental and practical significance to its application. Deep Neural Networks (DNNs) demonstrate significant potential in autonomously learning the interactions between high-dimensional inputs. This study proposed a deep neural network-based method for optimizing chemical reaction mechanism parameters, producing an optimized mechanism file as the final output. The novelty lies in two aspects: first, it systematically compares three DNN structures (Multi-layer perceptron (MLP), Convolutional Neural Network, and Residual Regression Neural Network (ResNet)) with other machine learning models (generalized linear regression (GLR), support vector machine (SVM), random forest (RF)) to identify the most effective structure for mapping combustion-related variables; second, it develops a ResNet-based surrogate model for ammonia–hydrogen mechanism optimization. For the test set (20% of the total dataset), the ResNet outperformed all other ML models and empirical correlations, achieving a coefficient of determination (R²) of 0.9923 and root mean square error (RMSE) of 135. The surrogate model uses the trained ResNet to optimize mechanism parameters based on a Stagni mechanism by mapping the initial conditions to experimental IDT. The results show that the optimized mechanism improves the prediction accuracy on laminar flame speed (LFS) by approximately 36.6% compared to the original mechanism. This method, while initially applied to the optimization of an ammonia–hydrogen combustion mechanism, can potentially be adapted to optimize mechanisms for other fuels.

Methanol is a liquid fuel with high oxygen content and the potential for a closed-loop carbon-neutral production cycle. To investigate the mixture formation and combustion characteristics of a heavy-duty Port Fuel Injection (PFI) methanol engine, a three-dimensional numerical simulation model was established using the CONVERGE 3.0 software. Multi-cycle simulations were performed to analyze the influence of wall film dynamics on engine performance. The results indicate that the “adhesion–evaporation” equilibrium of the intake port wall film determines the in-cylinder mixture concentration. Due to the high latent heat of vaporization of methanol, severe wall-wetting occurs during the initial cycles, causing the actual fuel intake to lag behind the injection and leading to an overly lean mixture and misfire. Regarding injection strategies, the open valve injection (OVI) strategy utilizes high-speed intake airflow to reduce wall adhesion and improve fuel transport efficiency compared to closed valve injection. OVI refers to the fuel injection strategy that injects fuel into the intake port during the intake valve opening phase. The open valve injection strategy (e.g., SOI −500° CA) demonstrates distinct superiority over closed valve strategies (SOI −200°/−100° CA), achieving a 75% reduction in wall film mass. The long injection duration and early phasing allow the high-speed intake airflow to carry fuel directly into the cylinder, significantly minimizing wall film accumulation and avoiding the “fuel starvation” observed in closed-valve strategies. Additionally, OVI fully utilizes methanol’s latent heat to generate an intake cooling effect, which lowers the in-cylinder temperature and helps suppress knock. Furthermore, a dual-injector strategy is proposed to balance spatial atomization and rapid fuel transport, which achieves a 66.7% increase in the fuel amount entering the cylinder compared with the original strategy. This configuration effectively resolves the fuel induction lag, achieving stable combustion starting from the first cycle.