| The calibration of traditional engine is mainly done by manpower,with long calibration time and high resource consumption,which makes it difficult to adapt to the development of intelligent engine control.Therefore,in recent years,the engine virtual calibration technology based on high-precision values and hardware-in-the-loop testing has gradually attracted the attention of the industry.The high-precision virtual engine can be used to replace the actual bench for calibration,which reduces the calibration workload of the bench and shortens the development cycle.However,virtual calibration has high requirements on the accuracy of the model and model accuracy depends on data.If a large number of experimental data has to be obtained,virtual calibration will be meaningless.In order to overcome the contradiction between limited data and model accuracy,this study builds engine virtual calibration technology by doing research on existing knowledge and prior experience,experimental design and engine modeling method.This research establishes a high-precision engine modeling framework for virtual calibration,which has functions on data analysis,model processing,model solving and so on.It improves the efficiency of pre-modeling preparation work.Then a virtual engine structure model was built based on GT-POWER for the commercially available dual VVT in-cylinder direct injection gasoline engine.On the basis of engine theoretical model,throttle coefficient model,injection characteristic coefficient model and semipredictive combustion model were built.According to the engine experimental data,based on D optimal design,the sample points of the throttle valve model,fuel injection characteristic model and semi-predictive combustion model were selected respectively.The accuracy of the model can make the fitting coefficient R2 of throttle valve flow coefficient,fuel injection pulse,CA50 and combustion duration reach 0.98,0.99,0.95 and 0.88.By analyzing the distribution law of sample points,the general scheme of experimental design is summarized and verified by another direct injection gasoline engine.The fitting coefficients R2 are 0.99,0.99,0.92 and 0.86 respectively,which proves that the experimental design scheme is universal.The engine model is a complex,coupled and nonlinear system.Even if the accuracy of the sub-models has reached a high level,due to the limitations of onedimensional numerical simulation and the inevitable errors in the modeling process,the accuracy of the overall engine model still may not meet the needs of virtual calibration.Aiming at the solving these challenges in engine modeling,this study established a multi-parameter synchronous optimization platform to optimize the correction factors in the model based on the NSGA-Ⅱ optimization algorithm,thereby improving the accuracy of the model.The accuracy of the virtual engine is verified based on the engine experimental data,which shows that the method developed in this study,using the model established by 96 operating conditions within the range of 2600 experimental points required by the traditional calibration,The proportion of the prediction error within 5% of intake manifold pressure,airmass,IMEP,temperature before turbine and pressure before turbine are 94.5%,90.5%,87%,88% and 91% respectively,which shows that the virtual engine model can carry out initial calibration for engine maps。... |
PDF Full Text Request