Study On Combined Prediction Of Gasoline Engine Ignition Advance Angle Based On Neural Network And LSSVM

With the rapid development of society,fossil fuels are drying up,governments have enacted more and more stringent regulations on motor vehicle emissions.Therefore,improving the matching accuracy of each parameter in the operation of the gasoline engine is the only way to reduce emissions and improve performance.The ignition advance angle of gasoline engine is an important parameter which can directly affect the performance of the automobile.If the ignition advance angle is too large,the engine is prone to knock and efficiency is reduced.On the contrary,the ignition advance angle is too late,the fuel consumption increases and the dynamic variation.With the development of the intelligent control algorithm,the fluctuation of AFR of the gasoline engine has been controlled at about 3%,while the ignition advance angle is still determined by the traditional look-up table,which is difficult to accurately control the ignition advance angle of the engine under all operating conditions.Due to the complex relationship between the operating condition and the ignition advance angle,it is necessary to obtain the ignition advance angle accurately by the experiment.However,the operating condition of the gasoline engine is a great many,so it is impossible to calibrate all the operating conditions.Therefore,it is of great theoretical significance and engineering application value to carry out the research on the prediction of ignition advance angle.This paper takes the ignition advance angle control of gasoline engine as the main research route.First of all,it introduced the impact of gasoline engine ignition timing on engine performance,as well as its precise control of the necessity and importance of the engine,and analyzed and studied the development of the ignition advance angle and the ignition advance angle control strategy.Secondly,the prediction model of ignition advance angle of gasoline engine was established based on BP neural network which is used to predict the ignition advance angle and the initial weights and thresholds of the BP network are optimized by the particle swarm optimization(PSO)algorithm based on the experimental data of ignition advance angle.Then,the prediction model of gasoline engine ignition advance angle based on LSSVM is established which initial parameters γ and σ2 are optimized by Cross validation method with the the same ignition angle of the sample data,and the prediction of ignition advance angle is simulated by LSSVM prediction model.Finally,after analyzing the defects of single model in predicting the ignition advance angle of gasoline engine,a variable optimal weighting method is proposed according to the variation of engine load and a variable optimal weighted combination model of BP-LSSVM ignition advance angle is established.By comparing and analyzing the results of other prediction model,the accuracy and validity of the BP-LSSVM ignition advance angle combination forecasting model based on variable weight is verified.