基于红外光谱-神经网络预测汽油辛烷值

PREDICTING GASOLINE OCTANE NUMBER WITH INFRARED SPECTRUM-ARTIFICIAL NEURAL NETWORK

根据汽油组分辛烷值与红外光谱峰面积分析数据,用人工神经网络(ANN)的反向传播(BP)算法建立了 汽油组分辛烷值神经网络预测模型,检验表明,ANN方法能准确地关联红外光谱分析数据与汽油组分辛烷值的关 系。马达法辛烷值与研究法辛烷值预测平均误差分别为0．192,0．178。

Based on the analysis data of gasoline octane number and infrared spectrum area, a neural network model is established to predict gasoline octane number by using the back-propagation （BP） algorithm of artificial neural network （ANN）. It is shown that an accurate correlation can be made between the infrared spectrum analysis data and gasoline octane number with ANN. The average prediction errors of MON and RON are 0.192 and 0.178 respectively.