小波神经网络的大坝变形预测研究

为了达到掌握大坝变形规律,确保大坝安全运行的目的.通过对小波神经网络和BP神经网络的对比,从隐含层激励函数的构造分析,得出两种网络本质相同,从BP算法的权值调节分析两种网络的预测性能包括收敛性能和泛化性能,并结合实践应用到具体的大坝预测问题上,验证小波神经网络在预测方面精度高,误差不超过0.1mm,同时泛化性能好的优势.

Research on the Predication of Dam Deformation Based on Wavelet Neural Network

The study aims to learn the dam deformation patterns to ensure dams’ safety .The comparison between the wavelet neural network and the BP neural network and the analysis of the structures of the hidden layer incentive functions show that both networks are same in essence .The predicting performance of both networks ,including convergence and generalization performance ,is analyzed from the value adjustment of BP algorithm .The application of the algorithm in a practical dam prediction verifies that the wavelet neural network is of good generalization performance ,and of high precision in prediction with the error less than 0 .1 mm .