BP神经网络的改进算法及在大坝变形分析与预报中的应用

Modified Algorithm of BP Neural Network and Its Application on Analysis and Prediction of Dam Deformation

针对传统的BP算法收敛速度慢、精度较低的缺陷,本文以西津大坝26#点的50期观测数据为例,选择了4个变形因子,进行了改进的BP神经网络在大坝变形分析与预报中的应用与研究.在此基础上,比较了改进的BP神经网络与传统的BP神经网络的预报效果.结果表明:改进的BP神经网络无论在学习算法的收敛速度,还是在预报精度上,都得到了大幅度的提高,且有一定的推广意义.

In view of the deficiency of standard BP neural networks slow training speed and low precision, using the monitoring data of Xijin Dam, choosing 4 factors, analysis and prediction of dam deformation based on modified BP neural network is studied. On the basis, prediction effect of the modified BP neural network and standard BP neural network is compared. The results show that： the modified BP neural network is practical significant to be extended on training speed and prediction precision.