人工神经网络用于近红外光谱预测汽油辛烷值

Application of Artificial Neural Network for the Prediction of Gasoline Octane Number by Near-infrared Spectroscopy

本文对BP人工神经网络(ANN)方法在汽油的辛烷值与其近红外光谱光谱吸光度的关系之间进行关联预报方面进行了研究。采用35个汽油实际样本数据,建立了利用汽油的近红外光谱光谱吸光度预测汽油辛烷值的BP人工神经网络模型。对所有辛烷值的计算结果与实测值进行了比较,得到的预测值与实测值计算误差小于1.55%。

The relation between the gasoline octane number and the near-infrared absorbence of gasoline by using BP artificial neural network for the prediction of gasoline octane number was studied in this article. 35 near-infrared spectroscopy data, were used to construct the BP artificial neural network model for the prediction of gasoline octane number. All the predicted values by the model have been compared with the practical values, and the error was less than 1.55%.