Research On Emission Prediction Modeling And Multi-objective Optimization Of Diesel Engine Fueled With Biodiesel

Under the dual pressure of energy crisis and environmental pollution,biodiesel has become one of the most potential alternative fuels for diesel engine because of its clean and renewable characteristics,high similarity of physical and chemical properties with fossil diesel.Therefore,the emission control of diesel engine fueled with biodiesel has become a research hotspot in the field of internal combustion engine.In view of this,the emission prediction model of diesel engine fueled with biodiesel was established by using machine learning algorithm such as BP neural network(BP-ANN)and support vector machine(SVM)in current work.The intelligent optimization algorithms such as genetic algorithm(GA)and particle swarm optimization(PSO)were employed to optimize the prediction accuracy of the established model.In addition,the optimized emission prediction model based on support vector machine(SVM)was combined with non-dominated sorting genetic algorithm(NSGA-II)to establish a multi-objective optimization model of pollutant emissions from diesel engine fueled with biodiesel.The optimal solution of biodiesel physical and chemical properties was obtained by using this multi-objective optimization model,which could reduce simultaneously the NOx and particulate matter emissions.The research results can provide theoretical basis and technical means for the optimization and regulation of physicochemical properties and components of biodiesel.The main research results obtained in current work are as follows:(1)Based on the experimental data of diesel engine benches,the emission prediction models of diesel engine fueled biodiesel were established by the BP neural network(BP-ANN)and SVM algorithm.The effects of neuron number on the accuracy of the BP-ANN prediction model,the effects of kernel function,penalty factor(C)and kernel function parameter(g)on the accuracy of SVM prediction model were studied.In the BP-ANN emission prediction model,the highest prediction accuracy could be achieved by respectively using 6 and 9neuron of hidden layer in the NOx emission prediction and particulate matter emission prediction.The determination coefficient(R~2)reached up to 0.9626 and 0.966 in the NOx emission prediction and particulate matter emission prediction respectively.In the SVM emission prediction model,highest prediction accuracy could be achieved by employing the radial basis kernel function.The determination coefficients(R~2)of NOx and particulate matter emission prediction reached up to 0.9555 and 09565 in the test set respectively,the mean square error(mse)of NOx and particulate matter emission prediction reached up to 0.00932and 0.00912 in the test set respectively.Moreover,the penalty factor(C)and the kernel function parameter(g)significantly influenced the prediction accuracy of SVM emission prediction model.Random C and g obviously decreased the determination coefficients(R~2<0.41)and increased the mean square error(mse>0.2)in the NOx and particulate matter emission prediction.(2)The weight and threshold value of BP-ANN prediction model,penalty factor C and kernel function parameter g of SVM prediction model was optimized by GA algorithm to establish GA-BP and GA-SVM prediction models respectively.Compared with GA-BP prediction model,GA-SVM prediction model has higher prediction accuracy.In NOx and particulate matter emission prediction,the R2 was increased by 1.8%and 2%respectively,and mse was decreased by 0.09%and 0.11%respectively.In addition,The SVM prediction model was also optimized by PSO algorithm to establish the PSO-SVM prediction model.By comparison of the prediction performance between GA-SVM and PSO-SVM prediction models,it was found that the two models had comparable prediction accuracy.However,the prediction speed of GA-SVM prediction model is faster.(3)The GA-SVM emission prediction model was combined with NSGA-II algorithm to establish a multi-objective optimization model of pollutant emissions from diesel engine fueled with biodiesel.The optimal solution of biodiesel physical and chemical properties was solved for NOx and particulate matter emissions.The results show that the NOx and particulate matter emissions from diesel engine could be reduced simultaneously by changing the components of biodiesel to adjust its physical and chemical properties.