基于小波分析-PSO优化ANFIS的径流组合预测模型研究  

Runoff Forecast Based on Wavelet Analysis and PSO Optimized ANFIS' Combined Model

作　　者：林武星 王玉明 范力 

机构地区：[1]中国葛洲坝国际工程有限公司,北京100022

出　　处：《中国农村水利水电》2011年第5期21-23,27,共4页China Rural Water and Hydropower

摘　　要：径流序列由于多种因素的影响,存在十分复杂的变化特性,如果利用常规方法直接进行预测,其预测精度一般不高。如果将径流序列分解成比较简单的序列,再利用非线性预测方法进行预测,其精度将会得到提高。利用小波分析的多分辨率分解功能对径流序列进行分解,降低了径流序列的复杂程度。利用PSO对自适应神经模糊推理系统的网络参数进行优化,提高模型结构参数的确定精度。将小波分析与PSO优化的ANFIS进行组合对径流序列进行预测,经实例验证:该模型能够提高径流预测的精度,预测效果较好。Affected by a variety of factors,runoff series are charactered by very complex change characteristics.If conventional methods are used to predict them directly,their accuracy is seldom very high.If we decompose the runoff series into more simple series,we will use nonlinear prediction methods to predict it and its accuracy will be improved.In this paper,wavelet analytical multi-resolution decomposition function is used to decompose the runoff sequences,it can reduce runoff series＇ complexity factor.PSO is used to optimize the network parameters of adaptive neuro-fuzzy inference system.It can improve the precision determination of structure parameters.It uses wavelet analysis and PSO optimization ANFIS combination model to forecast runoffs.The example shows that the model can improve the accuracy of runoff prediction,the prediction result is better.

关 键 词：径流序列预测 多分辨率分析 ANFIS PSO 组合模型 

分 类 号：TV121[水利工程—水文学及水资源]