摘要
利用人工神经网络理论对均匀加热垂直上升圆管内的临界热流密度(CHF)进行预测和参数趋势分析。本研究采用局部条件假设,并选用Groenevld的CHF查询表数据为本文神经网络训练的样本,采用训练成功的网络预测CHF值可以得到比常规方法更好的效果,其均方差为14.9%。
The critical heat flux (CHF) are predicted and its parametric trends are analyzed by applyinartificial neural networks (ANNs) to the CHF data base of upward flow water in uniformly heated vertical round tubes.The prediction and analysis are based on the local conditins hypotesis.Groeneveld's CHF Look_up Table is used to train the ANNs,and the trained ANN predicts the CHF better than any other conventional correlations method,with root mean square (RMS) error of 14%.
出处
《核动力工程》
EI
CAS
CSCD
北大核心
1999年第2期182-185,共4页
Nuclear Power Engineering
基金
空泡物理与自然循环重点实验室基金
关键词
神经网络
临界热流密度
参数趋势分析
反应堆
Artificial neural network Critical heat flux parametric trends analysis