Intelligent Calibration And Application Of Engine Performance Simulation Numerical Model Based On General Trend

Under the threat of energy crisis and environmental pollution,most countries have successively introduced a series of policies to limit energy consumption and emissions of vehicle.But at this stage,the battery capacity of pure electric vehicles cannot fully meet the requirements of users.Combining the internal combustion engine and the electric motor as the power system of hybrid electric vehi cle takes into account the requirements of economy,emissions and cruising range.With proper design and control in hybrid power system,Atkinson cycle engine can be significantly more economical than Otto cycle.In order to optimize the economic performan ce and emission performance of an Atkinson cycle gasoline engine,this paper completes the intelligent calibration and intelligent optimization of the engine simulation model.First,the GT-Power physical model is established based on the actual data of the engine,and the NSGA II algorithm is used to calibrate the physical model based on the experimental data.Secondly,by changing the engine control parameters SA,VVT-I,VVT-E and EGR,the performance parameters BSFC,NOx and CO ₂ of the Atkinson cycle gasoline engine are optimized.In the fitness calculation of NSGA III,the SVM model is used to predict the performance parameters,which can greatly reduce the time required for optimization.In order to ensure the accuracy of the S VM model,the results of the physical model operation are used to enhance training SVM model.Finally,the optimal parameters obtained by the SVM model driven by the NSGA III algorithm are input into the physical model to obtain the optimal results.In this paper,the performance of two empirical models,BP and SVM,in predicting the performance of an Atkinson cycle gasoline engine is compared.The main conclusions of this paper are as follows.（1）A high-precision engine performance simulation model was de veloped and calibrated based on experimental data.The results show that the relative error of the important parameters between the simulated and experimental values is within 3%.It can be considered that the GT-Power physical model can accurately predict engine performance and emissions.（2）In the optimization process,the physical model only simulated 79 cases.Theoretically,the time cost of the method in this paper can be as low as 1/23 of the method for optimization algorithms to directly drive physi cal models（3）The MAPE of BSFC,NOx and CO₂ are 0.20%,5.78%,and 0.06%respectively,and the MAE of KI is 16.07.The largest MAPE belongs to NOx,but SVM’s prediction of the trend of NOx is accurate.（4）In three sets of optimization plans,plan A has the largest decrease in fuel consumption,about 7.07%;plan A has the largest decreas e in CO₂ and CO,about 6.62%and 5.50%,respectively;the NOx of plan A was reduced by 24.18%.The EER and EEE of plan A increased by 6.21%and 2.26%,respectively.In this paper,the NSGA II algorithm is used to calibrate the physical model based on the experimental data,and the NSGA III algorithm is used to drive the SVM model to guide the optimization of the physical model.The method in this paper can greatly improve the efficiency of engine model calibration and optimization,and has important engineering application and theoretical value.