Basic Calibration Research Based On Steady-state Working Conditions Of Gasoline Engines

With the development of the automotive industry,modern engine design has become more complex,and the number of electronic control parameters included in it has increased.These parameters directly affect the performance and economy of the engine.The traditional calibration method is no longer suitable for modern engines due to its time-consuming and error-prone.Therefore,the automation of calibration have become the inevitable trend of engine calibration research.With the increasing number of electronic control parameters and the increasing coupling between parameters,manual calibration is no longer suitable for modern engines.Modelbased calibration technology has become the mainstream calibration method due to its high efficiency,accuracy,and strong adaptability.This article takes a three-cylinder gasoline engine as the research object,with the goal of optimizing its economy and emissions,and studies the model-based calibration method.The main research work is as follows:(1)An basic calibration problem of the engine is made,and the cycle of the engine is represented as the intake,ignition,and fuel injection operations.The control parameters that need to be calibrated are focused on the charging efficiency,air-fuel ratio,intake(exhaust)VVT,and ignition angle parameters.By comparing different experimental design methods,a space-filling design is used to develop an experimental design plan and collect experimental data.(2)To solve the problems of traditional calibration depending on test benches,long cycles,and high costs,this paper establishes an engine performance and emissions prediction model based on Gaussian Process Regression algorithm.The data collected based on the experimental design plan is modeled using the GPR algorithm.By comparing the advantages and disadvantages of various covariance functions,the Gaussian kernel function is selected as the covariance function,and the Particle Swarm Optimization algorithm is used to obtain the optimal hyperparameters of the GPR model.The model is tested using test data,and the results show that the PSO-GPR model has high prediction accuracy and can be used for subsequent calibration optimization.(3)To optimize the basic control parameters of the engine,this paper uses the PSO-GPR model constructed to calibrate the basic control parameters using the Non-Dominated Sorting Genetic Algorithm-II(NSGA-II)based on the elite strategy.The BSFC and NOx are used as optimization objectives of NSGA-II,and CO emissions,HC emissions,and particulate matter emissions are used as constraints.Multi-objective optimization is carried out,and each operating point will get a set of Pareto solutions.(4)182 operating points were set based on engine operating conditions,and all these points were optimized while ensuring a smooth map.Control parameter values were selected point by point to generate an initial map,which was then tightly smoothed using linear interpolation.The optimized map was imported into the ECU and tested on a test bench,and the results showed that the optimized engine had an expanded economic region and a reduced emission region.To address the issue of lengthy point-by-point calibration,this paper combined the results of the NSGA-II algorithm with a method for generating smooth maps.For the WLTC cycle,the optimization results of reference points based on time weighting and key point optimization were used to generate multiple maps,and the performance and smoothness of these maps were evaluated to automatically select some excellent calibration maps.In addition,based on some experimental data of the gasoline engine,we verified the results of this method in the GT-POWER simulation environment.The simulation results showed that the method for generating smooth maps can achieve rapid calibration and produce effective results.Calibration engineers can choose rapid or point-by-point calibration according to their needs.