神经网络在水位预报中的研究

STUDY AND APPLICATION ON INTERPRETATION METHOD OF AREA RAINFALL BASED ON NEURAL NETWORK

根据各站点降水资料,运用高斯权重客观分析原理计算广西区域的面雨量,以上游面雨量值、水位值与西江流域的梧州水位进行相关分析,找出基于水位的主要预报因子;进一步采用人工神经网络与主成分分析(PCA)相结合的方法进行了西江流域梧州水位的预报方法研究。计算结果表明,该预报方法所构造的预报模型对历史样本拟合精度高,试报效果也较好,可在预报业务中应用。

In terms of station precipitation data and Guass weight objective analysis theory, the area rainfall over Guangxi is calculated. A correlative analysis is made between area rainfall and water level value over the upper region and water level at Wuzhou over Xijiang valley. Based on this,main forecast factor of water level is selected. Furthermore, combined artificial neural network（ANN） method and principal component analysis（PCA） method, the study on forecast method of water level at Wuzhou over Xijiang valley is made. The results show that the prediction model, which is established by the method, has high fitting accuracy on historical samples and good testing effect as well, and can be applied to forecast operation.