基于小波分析和模糊神经网络的水文预测

Study on Hydrological Forecasting Based on Wavelet-fuzzy Neural Networks

提出一种基于小波 -模糊神经网络的水文时间序列预测方法 .利用小波分析具有“数学显微镜”的特点 ,分析水文时间序列的频率构成 ;通过模糊逻辑和神经网络两种理论的融合 ,对各频率分量进行预测 ,最后合成预测结果 .对浙江源口水库 1 0年间入库水量时间序列的预测实践 ,验证了方法的有效性 .

The paper brings out a forecasting method for the hydrological time-series based on wavelet-fuzzy neural networks.Various frequency signals are acquired by wavelet analysis which has “math microscope” characteristics.Various frequency signals are modeled through adaptive networks based on fuzzy inference system which integrates fuzzy logic theory and neural networks theory.Finally various signal forecasting results are added.Study on the modeling of hydrological time-series for Zhejiang Yuankou Reservoir approves the validity of the method.