基于两种神经网络的管内沸腾汽液两相流型识别  

Identification of the Boiling Gas-Liquid Two-Phase Flow in Tubes Based on Two Kinds of Neural Networks

作　　者：陈雅群[1] 郭雷[1] 张树生[1] 程林[1] 

机构地区：[1]山东大学

出　　处：《化工机械》2010年第6期763-766,共4页Chemical Engineering & Machinery

基　　金：国家重点基础研究发展计划(973)项目(No.2007CB206900)资助

摘　　要：将Fluent数值模拟得到的温度、压力和截面含气率等参数作为流型识别的特征向量,输入经过训练的BP及Elman神经网络,进行流型的识别。结果表明,在所取预测组数范围内,BP神经网络的识别准确率为95.24%,Elman神经网络对各种流型的识别准确率为100%。The temperatures,pressures,void fractions,and other parameters obtained from Fluent numerical simulation were used as the feature vectors of flow pattern identification,and the trained BP and Elman neural networks were inputed to identify the flow patterns.The result shows that the identification accurate rate of BP neural networks was 95.24%,and that of Elman neural networks identifying various flow patterns was 100%.

关 键 词：汽液两相流 ELMAN神经网络 BP神经网络 流型识别 

分 类 号：TQ021.2[化学工程]