基于BP神经网络的汽油抗爆剂敏感性预测

The Sensitivity of Gasoline Antiknock Forecast Based on BP Neural Network

研究了基于BP神经网络的建模方法,将其应用于93号汽油调合系统中研究法辛烷值的预测,考虑了93号高牌号汽油调合生产中组份比例、MMT加入比例、罐底油量、满罐油量、罐底辛烷值和空白辛烷值对成品辛烷值的影响,利用该模型对兰州石化公司实测值进行训练和测试.应用结果表明,该模型的预测精度完全能达到工业生产的要求,基本合理可靠.

Modeling method based on BP neural network is studied,in which the No. 93 high--grade gas- oline is applied to the gasoline blending system in the forecast of octane number. It considered the No. 93 high--grade gasoline blending components in the production ratio, of the Join ratio MMT, the bottom of oil cans, full fuel tank, the tank with the bottom and the blank octane number impact on the finished octane number. This model is used in training and testing for Lanzhou Petrochemical Company procee- ded, the result shows that the forecast accuracy of the model achieves the requirements of industrial production, which is reasonable and reliable.