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%.
出处
《化工机械》
CAS
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
关键词
汽液两相流
ELMAN神经网络
BP神经网络
流型识别
Gas-Liquid Two Phase Flow
Elman Neural Network
BP Neural Network
Flow Pattern Identification