The BP Neural Network Based On Improved Ethanol And Gasoline Emission Analysis

The application of the engine has been constantly development many years, especially in the auto possession is increasing year, engine emissions have become the focus of research, ethanol fuel as biomass fuel, our country also constantly expand to new fuel-ethanol fuel application. But ethanol fuel compared to regular gas emissions advantages and disadvantages problems are still in a fuzzy condition, A large number of experimental work for these research of this aspect, and experiments are based on the complex expensive experiment equipment. It also has a lot of research using mathematical modeling method to engine emission characteristics of calculus, but the engine working process is a complicated process, the mechanism of the engine combustion emissions and internal structure parameters have not yet quantification. Based on the above two reasons, this paper puts forward the neural network modeling methods in experimental equipment ordinary, experimental data quantity is little, model calculus not complex situations steady state and forecast engine transient emission performance. The paper mainly studies for content.According to the engine condition, Emission characteristics based on the BP neural network of engine emission forecasting model. The training sample selection methods is variable boundary orthogonal experiment, Hidden neurons structure is optimized by using grey correlation analysis method Model is established for the steady-state condition after engine emission experiments, engine burn ethanol (E10) and the common gasoline and work for load condition experiment, this paper builds the steady-state engine emission forecasting model.According to the neural network excellent fan capacity, this article establishing the engine steady-state emissions prediction model can also predict the sample which does not appear. Model by adding to cool water temperature for new samples for transition sample forecast15condition combination mode of the car emissions, and can show ethanol for gasoline and regular gasoline in of the discharging characteristics of the different working conditions. Prediction results show that engine emission performance follow work cycle rule.This article finds a new direction modeling and design idea for the engine emission performance, it also added certain theoretical and practical basis for the emission regulations, and the methods in this paper have certain application value.