ET_0的主因子和主成分神经网络模型比较  

Comparison of Principal Factors and Principal Components Neural Network Model of ET_0

作　　者：赵璐[1] 蔡焕杰[1] 王健[1] 

机构地区：[1]西北农林科技大学旱区农业水土工程教育部重点实验室,陕西杨凌712100

出　　处：《节水灌溉》2010年第3期23-25,29,共4页Water Saving Irrigation

基　　金：国家"863"计划项目(2006AA100202);国家科技支撑计划(2007BAD88B10)

摘　　要：为了简化BP神经网络预测ET0的模型,将气象因子包括最高、最低和日平均温度、日照时数、相对湿度和风速进行主成分分析和偏相关分析,提取主成分和主因子,分别建立了基于主成分和主因子的三层BP神经网络模型,并对两种模型的训练和预测结果进行比较。选取湟水流域的乐都气象站2003年到2006年5月逐日的气象资料,采用Matlab神经网络工具箱进行模型训练与预测。结果表明主成分神经网络训练和预测模型的精度都优于主因子神经网络模型。主要是由于两种模型选取输入层的因子不同造成的。In order to simplify neural network model for the prediction of ET0, principal component analysis and partial correlation a- nalysis are applied to the weather data, including the maximum, minimum and average daily temperature, sunshine, relative humidi- ty and wind velocity. And a three-layer BP（back-propagation）neural network model is constructed based on the principal components and principal factors, respectively. The training and prediction results are compared. Based on daily meteorological dates of May from 2003 to 2006 in LeDu Station in Huang river basin, the two models were trained and predicted with Matlab neural network tool- Box. The results show that the principal-component-based BP network model is superior to the principal-factor-based BP network model in the training and prediction of ETo. The main reason is the different factors of input layer in two models.

关 键 词：ET0 主成分分析 偏相关分析 神经网络 

分 类 号：S161.4[农业科学—农业气象学]