PLS分析与RBF神经网络耦合环境模型  

Environmental Coupling Model of PLS Analysis and RBF Neural Network

作　　者：龚敏庆[1,2] 孙明伟[2] 金明仲[1] 

机构地区：[1]贵州民族学院理学院,贵阳550025 [2]贵州大学理学院,贵阳550025

出　　处：《数理统计与管理》2011年第5期840-845,共6页Journal of Applied Statistics and Management

基　　金：国家统计局基金(2009LZ009);国家民委基金(10GZ08);贵州科技厅基金(黔科合外G字[2010]7011);贵州省自然科学基金(黔科合J字[2010]2136);贵州省2010年度省长资金;毕节行署贵大基金(毕循专合字[2010]SK003)

摘　　要：鉴于城市大气环境质量受到诸多复杂因素影响,且各因素间存在多重相关性,本文将偏最小二乘(PLS)分析与人工神经网络径向基网络(RBF)耦合,建立偏最小二乘径向基神经网络模型(PLSRBF),应用于贵阳大气环境质量的检验和预测。实例表明:PLSRBF模型可对原多自变量模型进行降维简化,并可有效提取解释变量信息,防止信息丢失,且具有较强的拟合能力。As urban atmosphere environmental quality was influenced by various complicated factors among which there was multiple correlation, a Partial Least Square Radial Basis Function（PLSRBF） artificial neural network model which was used to test and predict atmosphere environmental quality of Guiyang was proposed by coupling partial least square analysis and RBF neural network. The results showed that the PLSRBF model could simplify the original multiple variables model via reducing its dimension, and extracted information effectively from independent variables avoiding losing information, which was of high fitting ability.

关 键 词：大气环境质量 偏最小二乘分析 RBF神经网络 拟合预测 

分 类 号：O212[理学—概率论与数理统计]