神经网络方法在原油评价数据库物性计算中的应用

Neural Network Application in Calculation of Physical Properties in Crude Oil Evaluation Database

在原油评价数据库中,为提高物性计算的精度,采用神经网络方法计算原油中馏分油物性。用此方法可以计算汽油的辛烷值、柴油的十六烷值、馏分油中芳烃和硫含量。该方法的计算误差小于2％。用此方法进行物性计算,提高了计算精度,此方法也可以应用到其他原油物性计算之中。

In crude oil evaluation database, physical properties are calculated by using a neural network method. Through it, octane value of gasoline, cetane value of diesel oil, aromatics and sulphur contents in fraction oil can be calculated. The deviations between calculated values and measured values are less than 2% . The calculation accuracy has been improved by using such a method, and which is also appropriate to be applied for calculation of other crude oil properties.