时间序列小波神经网络组合模型在年降雨量预测中的应用  

Annual Rainfall Prediction Based on AR-WNN Model

作　　者：陶猛[1] 徐淑琴[1] 

机构地区：[1]东北农业大学水利与建筑学院,哈尔滨150030

出　　处：《节水灌溉》2013年第12期40-42,45,共4页Water Saving Irrigation

基　　金：黑龙江省教育厅科研项目<黑龙江省灌区水资源动态预测方法与应用研究>(11551044)

摘　　要：在研究了降雨量特性的基础上提出了一种新型的组合预测模型即时间序列-小波神经网络模型(ARWNN)。以858农场为例,先利用时间序列模型对降雨量进行拟合,并将拟合结果作为小波神经网络的输入进行建模分析,并建立时间序列AR模型和小波神经网络模型与之比较,通过精度检验和对比分析结果表明,组合模型对降雨量的拟合及预测精度均较高。Based on the research of tt^e rainfall characterlst^cs, a time-series w^tv~ x combined model （AR-WNN）. Taking the 858 Farm as an example, the time series model is used to simulate the rainfall and the re- sult is used as the input of wavelet neural network model to establish a new model. It is compared with the time-series AR model a^d the wavelet neural network model. The precision examination and contrast analysis result indicates that the combined model has high- er precision for the simulation and prediction of rainfall.

关 键 词：降雨量 AR—WNN AR模型 小波神经网络模型 

分 类 号：TV125[水利工程—水文学及水资源]