基于BP神经网络的渠道墩柱断面过冰能力预测

Prediction of Ice Discharge Capacity of Channel's Pier Column Section Based on BP Neural Network

冰凌问题是寒冷地区渠道冬季输水面临的一个主要问题,预测易出现冰塞的桥墩、闸墩等束窄断面的过冰能力,对冰水二相流安全输水具有重要意义。通过物理模型试验分析墩柱断面的渠道过冰能力,根据试验数据,采用量纲分析与BP神经网络方法,对渠道墩柱断面过冰能力进行预测,预测结果较单一BP神经网络预测和非线性拟合得出的结果更为精确。

The ice problem is a major problem in winter water delivery in cold area. Predicting icedischarge capacity of narrow sections where ice jam usually occurs has a significant mean to safety deliver ice-water two-phase flow. Through analyzing the ice discharge capacity of channel＇s pier column section by conducting physical experiment,according to the experiment data,combining dimensional analysis and BP neural network to predict ice discharge capacity of channel＇s pier column section. It is found that this method is much more accuracy than that of single BP neural network and nonlinear fitting method.