基于机器学习的分析方法，用于研究高效可再生甲醇燃烧的点火策略与机制：柴油引导式及湍流喷射点火发动机

《Fuel》：A Machine learning based analysis for ignition strategies and mechanisms of High-Efficiency renewable methanol combustion: Diesel-Piloted and turbulent jet ignition engines

低碳中和背景下甲醇发动机高效燃烧与排放控制研究，提出融合响应面法（RSM）、支持向量回归（SVM）及多目标粒子群优化（MOPSO）的新型智能分析方法。通过对比柴油辅助点火（DPDI）与湍流喷气点火（TJI）两种策略，发现DPDI模式下多阶段热释放使指示热效率达46.7%，但NOx排放较TJI高19.5%；TJI通过集中式点火实现更优排放控制，但对喷射时序敏感。研究揭示了两种策略的燃烧机理差异及关键参数（如预燃室能量比例、喷射间隔）的敏感性，为甲醇发动机点火系统优化提供理论依据。

　　
methanol combustion research presents a comprehensive analysis framework integrating multi-objective optimization and machine learning techniques to evaluate diesel-piloted and turbulent jet ignition (TJI) strategies. The study addresses critical challenges in low-carbon engine development by systematically comparing ignition methods for high-pressure direct injection (HPDI) methanol engines. Key findings reveal trade-offs between thermal efficiency and emission control that inform optimal strategy selection under varying operational conditions.

The research establishes a novel methodological approach combining response surface methodology (RSM) with support vector machine (SVM) regression and multi-objective particle swarm optimization (MOPSO). This framework effectively navigates complex parameter interactions while minimizing experimental requirements through fractional factorial design and model validation using analysis of variance (ANOVA). The integration of optimization algorithms with predictive models enables simultaneous evaluation of multiple performance metrics including indicated thermal efficiency (ITE), nitrogen oxides (NOx) emissions, and combustion stability.

Critical comparisons between diesel-piloted and TJI strategies emerge from the analysis. Diesel-piloted mode demonstrates superior thermal efficiency reaching 46.7% under optimized conditions through multi-stage heat release mechanism. This approach relies on distributed auto-ignition patterns created by pilot diesel injection, which provides multiple ignition points for robust combustion initiation. However, the dual-fuel system introduces operational complexity and results in 19.5% higher NOx emissions compared to TJI mode.

TJI strategy offers distinct advantages through centralized ignition generated by high-velocity turbulent jets. This configuration enables precise control over heat release timing and magnitude, resulting in significantly lower NOx emissions. The study quantifies performance differences across medium (75%) and high (100%) load conditions, revealing fundamental distinctions in combustion phasing and emission formation mechanisms. Sensitivity analysis identifies pilot energy proportion (EPp) and injection timing interval as critical parameters affecting strategy performance.

Methodological innovations include development of a predictive SVM model validated through ANOVA at 95% confidence level. This ensures reliable extrapolation beyond measured data ranges while maintaining model generalizability. The multi-objective optimization framework addresses conflicting requirements for efficiency and emissions reduction through Pareto front analysis, providing decision-makers with trade-off visualizations for optimal parameter selection.

Operational insights from the study reveal strategic advantages of each ignition method. Diesel-piloted systems maintain higher efficiency across load ranges but require careful balancing of pilot fuel quantity to avoid excessive NOx. TJI strategies achieve cleaner combustion with potential for ultra-lean operation but depend on precise jet dynamics and fuel injection synchronization. The research quantifies these relationships through empirical data fitting and parameter optimization, establishing a quantitative basis for engine design improvements.

Practical implications include recommended parameter ranges for different engine configurations. The study emphasizes that while diesel-piloted mode offers better thermal efficiency through multi-point ignition, TJI's centralized combustion provides superior emission control. This knowledge helps manufacturers decide between strategy selection based on specific performance priorities. Future development directions suggested include hybrid ignition systems combining benefits of both approaches, and advanced parameter tuning through adaptive control algorithms.

The research significantly advances methanol engine development by providing:
1. Unified evaluation framework for jet-assisted ignition strategies
2. Quantitative relationship between operational parameters and performance metrics
3. Optimized parameter combinations for specific application scenarios
4. Machine learning models for rapid system simulation and design iteration

Methodological contributions include:
- Development of RSM-based experimental design for complex systems
- SVM regression models validated through statistical significance testing
- Multi-objective optimization implementation for conflicting performance goals
- Sensitivity analysis revealing key parameter interactions

These advancements establish a new standard for comparing alternative ignition strategies in modern engine development. The framework's adaptability allows application to other fuel systems and combustion configurations, potentially reducing research costs and development cycles for low-carbon engine technologies. The study ultimately provides actionable guidelines for balancing efficiency and emissions in methanol-fueled engines through targeted parameter optimization and ignition strategy selection.