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.
出处
《节水灌溉》
北大核心
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)