基于人工神经网络模型预测常减压侧线收率  

Prediction of the Atmospheric and Vacuum Line Yield based on Artificial Neural Network Models

作　　者：岳鹏[1] 孙忠超[2] 刘熠斌[3] 

机构地区：[1]中海石油炼化有限责任公司,广东惠州516086 [2]山东三维石化工程股份有限公司,山东青岛266071 [3]中国石油大学(华东),山东青岛257061

出　　处：《广州化工》2015年第10期151-153,共3页GuangZhou Chemical Industry

摘　　要：基于人工神经网络模型对常减压侧线收率进行了预测,模拟计算采用Matlab软件建立径向基函数神经网络模型。在模拟计算的三十个数值结果中,二十五个数值的偏差控制在10%以内,其中九个数值的误差控制在3%以内,达到最初模拟目标。随着数据库的扩充,参与训练模型的数据增多,模型就更接近真实情况,也就更好的预测侧线收率。Yields of atmospheric and vacuum were predicted based on artificial neural network model, simulated by Matlab software which established radial basis function （RBF） neural network model. In thirty results of numerical simulation, 25 groups of numerical deviation were controlled within 10%, nine groups were in 3%, had completed the initial target. As the expansion of database, the number of training data increased, the model would reached the real situation, which could predict the line yield effectively.

关 键 词：人工神经网络 径向基函数 侧线收率 

分 类 号：TE624[石油与天然气工程—油气加工工程]