论文部分内容阅读
针对河套灌区地下水位预测问题,结合小波变换的时频局部特性和神经网络的逼近功能,构建了两种不同耦合方式下小波和BP神经网络相结合的小波网络模型,比较了不同耦合方式下小波网络模型与单纯神经网络模型的预测效果。两种耦合方式下的小波网络模型模拟结果均比单纯使用人工神经网络模型更接近实测值,对低频信号序列及高频信号序列分别进行神经网络模型预测后再进行重构的预测方式比直接将小波分解的多级信号与神经网络结合的预测方式具有更好的预测效果。
Aiming at the problem of groundwater level prediction in Hetao Irrigation District, a wavelet network model combining wavelet and BP neural network under different coupling modes is constructed by combining the time-frequency local characteristics of wavelet transform and the approximation function of neural network. Wavelets under different coupling modes are compared Prediction of network model and pure neural network model. Compared with the artificial neural network model, the results of the wavelet network model under the two coupling methods are closer to the measured values. The forecasting method of the neural network model after the prediction of the low-frequency signal sequence and the high-frequency signal sequence respectively is better than the direct Wavelet decomposition of multi-level signal and neural network prediction method with better prediction.