基于BP神经网络的柴油机掺烧生物柴油排放特性预测研究  

Emission Prediction of Diesel Engine Fuelled with Diesel/Bio-diesel Based on BP Neural Network

作　　者：李楠[1] 汤东[1] 李昌远[1] 徐亚超[1] 

机构地区：[1]江苏大学汽车与交通工程学院,江苏镇江212013

出　　处：《车用发动机》2013年第5期48-52,56,共6页Vehicle Engine

基　　金：国家自然基金项目(51176068);江苏高校优势学科建设工程资助项目(PAPD);江苏省2012国际合作项目(B22012025);江苏省2011研究生创新计划立项项目;南通市生物柴油重点实验室开放课题(S2011002)

摘　　要：运用BP神经网络相关理论,以折合燃油消耗率、负荷以及掺烧比为输入参数,以CO,HC,NOx和soot排放量为输出参数,建立了柴油机掺烧生物柴油排放的双隐层BP神经网络模型,并利用该模型对发动机不同工况和掺烧比下的排放物进行预测分析。结果表明:CO,HC,NOx和soot排放预测结果与试验值具有较好的一致性,各排放的期望响应与仿真输出的线性回归相关系数R很高;HC和CO排放预测值随折合燃油消耗率的升高而增加,NOx和soot排放预测值随折合燃油消耗率的增加而下降。Based on BP neural network relevant theories, the BP neural network model with double hidden layers for emission performance of diesel engine mixed with bio-diesel was established with the inputs of equivalent specific fuel consumption, load and mixing ratio and the outputs of CO, HC, NO, and soot emission. With the model, the pollutant emissions under different conditions and mixing ratios were predicted and analyzed. The results show that the predicted CO, HC, NO, and soot emission have a good agreement with their experimental values. The expected response and simulation output have high linear regression correlation coefficient R. With the increase of equivalent specific fuel consumption, the predicted values of HC and CO emis- sions increase and that of NOx and soot emissions decrease.

关 键 词：BP神经网络 生物柴油 排放 性能预测 

分 类 号：TK464[动力工程及工程热物理—动力机械及工程]