基于FCM的小波神经网络模型在径流预测中的应用

Wavelet neural networks model used for runoff forecast based on fuzzy C-means clustering

针对径流变化存在的季节性差异,提出了一种在对预报因子集进行模糊聚类分析基础上构建径流预测模型的新方法:先通过模糊C-均值聚类将历史径流数据进行分类,然后利用小波神经网络分别建立预报因子集与观测值之间的局部预测模型,文中采用网络模型分类识别器,可自动搜寻相适应的局部网络模型进行预测.以西南某水库2006年日平均入库来流的计算实例对简单小波神经网络预测模型和文中所建的融合模型进行了比较.

In consideration of the seasonal heterogeneity of runoff variety, a new method of runoff forecast based on.fuzzy clustering analysis for forecasting factor set is presented in this paper. The historical runoff data are divided into four categories by using fuzzy C-means clustering. Then local forecasting models between the forecasting factor set and observation values are established by using wavelet neural network model. And a categorized recognizer of network model is applied, which can automatically search a corresponding local model to forecast. A comparison between single wavelet neural model and the forecasting model proposed in this paper is made through an example. The results show that the forecasting precision of the latter is higher than that of the former.