水文时间序列的小波神经网络工具箱预测

Wavelet neural network toolbox prediction for hydrological time series

小波分析是傅里叶分析发展史上的里程碑式进展,具有同时揭示信号在时域和频域局部变化特征的能力,被誉为数学的"显微镜"。本文将小波分析与BP网络有机结合,提出一种小波神经网络工具箱预测水文序列的新方法,对水文时间序列进行趋势预测,并与传统的BP和RBF网络预测结果比较,实验表明小波网络工具箱预测模型对数据具有很强的适应能力,预测精确,潜在优势明显。

Wavelet analysis is a milestone in the development history of Fourier analysis.It has the ability to reveal the characteristics of local changes of signals in time domain and frequency domain simultaneously.In this paper,the organic combination of wavelet analysis and BP network, a wavelet neural network toolbox to predict new methods of hydrological time series is proposed, trend projections for hydrological time series, and compared with the traditional BP and RBF network forecast results, the experiments show that wavelet neural network toolbox prediction model has a strong ability to adapt to data, forecasting precision, potential advantage is obvious.