Research On In-cylinder Combustion Monitoring Method Of Marine Natural Gas Engine Based On Ion Current

With the increasingly stringent requirements of ship emission regulations,the future ship power has a trend of using natural gas as the main transportation fuel.Lean combustion can effectively improve engine thermal efficiency and reduce NOx emissions.Engines will face the problems of large cycle fluctuations and increased probability of partial combustion and misfire.Therefore,real-time monitoring and feedback control of the in-cylinder combustion process are required.The ion current method can easily transform the spark plug into a combustion state sensor,and has the potential to replace the expensive cylinder pressure sensor.Therefore,this subject tries to apply ion current technology in lean-burn natural gas engine,and uses artificial neural network to extract combustion information and other related parameters from ion current signal for feedback control.First of all,based on the single-point injection natural gas engine,this paper adds a high-energy ignition system and an ion current detection,acquisition and processing system to realize the detection of ion current and the extraction of characteristic parameters under lean burn conditions.In addition,the ion current detection is carried out in all common working conditions of the engine,and the factors and characteristics that affect the ion current are studied.Secondly,the correlation between the characteristic parameters of ion current and combustion parameters and emissions is analyzed.Finally,according to the characteristic that the thermal phase peak of the ion current will disappear under some working conditions,the training data is divided into two categories,and the two kinds of data are used to establish artificial neural network models with BP and RBF networks respectively.The study found that the increase of the excess air coefficient and the backward shift of the ignition timing will greatly reduce the intensity of the ion current signal,and the reduction degree of the thermal phase peak value is greater than that of the chemical phase peak value;after the excess air coefficient exceeds a certain value,the ion current thermal phase peak rapidly decrease or even disappear.The first peak phase of ion current has a poor correlation with the phase of the maximum pressure rise rate,but the first and second peak phases have strong correlation with other combustion phases.The correlation coefficients between the ionic current amplitude characteristics and the maximum combustion heat release rate and the maximum pressure rise rate are all higher than 0.8,and the correlation coefficient with the maximum in-cylinder pressure is between 0.6 and 0.7.The correlation coefficients of the three amplitude characteristics of ion current with NOx emissions are all higher than 0.8,and the correlation coefficients with other emissions are all lower than 0.8.The neural network model can accurately predict the excess air coefficient with an accuracy of more than 95%.In addition to the maximum pressure rise rate and its phase,the model can more accurately predict each combustion phase and parameters under different working conditions,with an error of less than 10%.Emission models can accurately predict NOx and CO emissions at higher concentrations,but lose accuracy at lower concentrations.The prediction accuracy of CH4 and HC emissions is low,with errors exceeding 10%.