基于人工神经网络(BP)方法预测汽油辛烷值

Prediction for Octane Number of Gasoline Based on Artificial Neural Network Approach Method.

本文基于人工神经网络（ＢＰ） 方法，用毛细管色谱法预测汽油馏分的辛烷值，其预测最大绝对误差为０．２８ ，平均误差为０．１２２，比常用的线性回归数学模型法更能准确地预报辛烷值。

Based on the Artificial Neural Network Approach method, the octane number of gasoline was predicted by using capillary chromatography. The maximum absolute error of prediction is 0.28, and the average error of prediction is 0.122 . The results show that the ANN approach method can predict the octane number of gasoline more accuracy compared with the common linear regression method.