人工神经网络在棒束临界热流密度预测中的应用

Application of Artificial Neural Network in Bundle Critical Heat Flux Prediction

摘要: 基于已有的棒束临界热流密度数据库，采用COBRA-Ⅳ程序计算得到子通道局部临界热流密度数据库。用人工神经网络（ANN）理论对数据库进行训练，得到基于ANN理论的棒束临界热流密度预测模型。预测模型的预测精度显著高于常用经验关系式的预测精度，其预测值的均方差为5.63%。

Abstract: A bundle critical heat flux(CHF) database based on subchannel local condition is obtained by analyzing existing bundle experimental database with COBRA-Ⅳcode. Artificial neural network is then applied to train the database and a bundle CHF prediction model is finally obtained. The prediction accuracy of the obtained model is much better than that from general empiric formula, and the root-mean-square of predicated value is 5.63%.