基于小波-ANFIS的水库年径流组合预报模型

Reservoir Annual Runoff ANFIS Forecast Model Based on Wavelet Analysis

提出了一种基于小波-ANFIS的水库年径流组合预报模型。利用Mallat算法对年径流序列进行多尺度分解,得到对应尺度下的低频信号和高频信号;分别对低频信号和高频信号建立ANFIS模型进行预报,将各模型预报结果叠加作为原径流的预报值。将该模型用于昭平台水库年径流预报中,通过与单一ANFIS模型预报结果的对比分析,验证了小波-ANFIS水库年径流预报模型的有效性。

This article applies wavelet-ANFIS model for reservoir annual runoff forecast . Annual runoff series data are decomposed into individual components at different scales by Mallat, and then low- frequency signals and high -frequency signals are obtained according to different scales. Based on different signals, ANFIS model is used to forecast low-frequency siganals and high-frequency siganals. Then the forecast results of every model are composed to form the forecast of original runoff. This model is used in Zhaopingtai Reservoir annual runoff forecast. By comparing this method with the single ANFIS forecast model and the model based on support vector machine, the feasibility of wavelet-ANFIS model for reservoir annual runoff forecast is proved.