In-cylinder Pressure Based Real-Time Prediction Technology Of In-Cycle Emissions Of Diesel Engines

Relying on combustion state closed-loop feedback control technology platform, thein-cylinder combustion process can be controlled and improved and thus the emissionclosed-loop control can be achieved through real-time prediction of emissions.Consequently the technology of real-time prediction of emissions in diesel engines hasgreat potential in terms of reducing emissions and improving economic efficiency. Richinformation of combustion process is contained in in-cylinder pressure signal, fromwhich combustion state parameters in strong correlation with emissions can be obtained.Thus the empirical emission model based on combustion state parameters can beestablished and written online codes to achieve real-time prediction of emissions. Thissubject was targeted on achieving real-time prediction of in-cylce emissions in bothsteady-state and transient operating conditions based on in-cylinder combustion analysistool. Specifically input parameters of NO_xand soot emission models were determined;offline empirical models of NO_xand soot emissions were built based on steady-state testdatas; real-time prediction of in-cylce NO_xemission was achieved based on thein-cylinder combustion analysis tool.Based on the mechanism of formation of NO_xand Soot, combustion stateparameters in strong correlation with NO_xand Soot emissions were extracted fromcurves of in-cylinder pressure, rate of in-cylinder pressure, rate of heat release andaccumulated heat release. Correlations between the burned zone maximumtemperature and NO_xemission, the formation area of high-heat NO and NO_xemission,the length of diffusive combustion and soot emission were analyzed particularly.Quantitative correlations between combustion state parameters and emissions werecalculated based on fixed operating conditions test datas. The conclusion was that themaximum pressure, maximum rate of pressure, lacation of maximum ROHR,location of50%total heat release, maximum burned zone temperature, formation areaof high-heat NO could be the inputs of the emipirical model of NO_xemission; themaximum pressure, maximum rate of pressure, lacation of maximum ROHR,location of50%total heat release, length of whole combustion, length of diffusivecombustion could be the inputs of the emipirical model of Soot emission.Steady-state modeling and validation tests were designed separately. Both methods of polynomial and exponent were selected for modeling. Models from combustion stateparameters to NO_xand Soot emissions were built based on modeling test datas. Theprediction accuracy of both methods were calculated, analyzed and compared based onvalidation test datas. Choosed accuracy of models and complexity of algorithm inin-cylinder combustion analysis tool as the performance indexes to compare theadvantages and disadvantages of both methods. According to the result of comparision,quadratic polynomial model was selected as the better suited model for real-timeprediction of diesel engine emissions.The in-cylinder combustion analysis tool was further developed based onMPC5644A MCU. New combustion state parameters and the quadratic polynomialmodel of NO_xemission were added in the orginal algorithm. Verified the accuracy ofreal-time prediction of in-cycle NO_xemission under the steady-state operatingconditions: except for one bad condition, prediction relative errors of the other15conditions were all less than10%; average prediction relative error of the16conditionswas6.8%. Verified the practicality of real-time prediction of in-cycle NO_xemissionaccording to the transient tests: real-time prediction of NO_xduring the transient processchanged slightly larger than the measured NO_x. Overall, during the transient processpredicted NO_xhad similar trend with measured NO_x.