Research On Laminar Flame Velocity And HCCI Combustion Characteristics In Gasoline Based On Machine Learning

As people pay more attention to environmental issues,emission regulations have become more stringent.The combustion of gasoline fuel in the transportation field is the main source of pollutant emissions.The combustion characteristics of gasoline fuel and the research on Homogeneous Charge Compression Ignition(HCCI)engines are of great significance for reducing pollutant emissions and improving combustion efficiency.However,the combustion of gasoline fuel,especially under new combustion conditions such as HCCI,is an extremely complex process,which is affected by many factors such as environmental factors,engine structure,fuel properties and so on.Therefore,there is an urgent need for a robust prediction model that can obtain the combustion characteristics of complex fuels under a wide range of engine operating conditions without the need for a large number of experimental measurements and calculations.To overcome these challenges,the thesis is based on the National Natural Science Foundation of China Youth Fund Project(Research on Gasoline Fuel Auto-ignition Characteristics and Multi-parameter Co-regulation Mechanism under Direct Injection Combustion Mode,Grant No.52006058),which developed a kinetic calculation-machine learning coupled model to investigate the influencing factors of laminar flame velocity and HCCI combustion phase as well as NOx emission.The research work and innovations of the thesis are as follows:(1)Four machine learning prediction models of laminar flame velocity are established.After comparing the four models,it was found that the GBDT model had the best performance in the test set,with RMSE of 0.0084 and R~2of 0.9984.(2)Based on the gradient boosting decision tree model,the influence of input parameters on laminar flame velocity was studied.The results show that temperature has a positive effect on laminar flame velocity and has a direct proportional relationship on logarithmic scale.The pressure can inhibit the laminar flame velocity,which is inversely proportional on the logarithmic scale.The distribution of laminar flame velocity on the ternary phase diagram of octane number-sensitivity-equivalence ratio shows three distinct regions:medium velocity region,high velocity region and chaos region.(3)A random forest classifier is established,and the complete combustion boundary of each research parameter is studied.The results show that the octane number-sensitivity-equivalence ratio coupled complete combustion boundary is enlarged with inlet temperature and inlet pressure,while the increase of exhaust gas recirculation(EGR)reduces the temperature-pressure-compression ratio coupled complete combustion boundary.(4)A multi-layer perceptron model was developed to analyses the effects of each input parameter on HCCI combustion timing,combustion duration and NO_xemissions.The study shows that with the increase of the equivalent ratio and compression ratio,the combustion timing is advanced and the combustion duration is shortened;the increase of EGR will cause the combustion timing to be shifted back and the combustion duration to be longer;the increase of octane number will cause the combustion timing to be shifted back,but the change in combustion duration is not significant;the increase of compression ratio,intake air temperature and equivalent ratio will all cause the increase of NO_xemission;while the increase of EGR will cause the reduction of NOx emission.The increase in compression ratio,intake temperature and equivalent ratio all result in an increase in NO_xemissions,while an increase in EGR results in a decrease in NO_xemissions.