偏最小二乘与人工神经网络耦合模型在酸雨pH值预测中的应用  

Partial least square method and BP neural network coupled model and its application to prediction of pH value of acid rains

作　　者：周秀平[1] 王文圣[1] 曾怀金 

机构地区：[1]四川大学水利水电学院,四川成都610065 [2]四川省水利水电勘测设计研究院,四川成都610074

出　　处：《水利水电科技进展》2006年第4期50-52,共3页Advances in Science and Technology of Water Resources

基　　金：四川省科技厅软科学基金资助项目(042R025-051)

摘　　要：将偏最小二乘与人工神经网络耦合,建立了一种新型耦合预测模型———偏最小二乘与人工神经网络耦合模型。该模型利用了偏最小二乘方法有效处理自变量之间多重相关性问题和人工神经网络可以较好地解决非线性问题的能力,在城市酸雨pH值预测中的应用表明,该模型预测精度高,明显优于偏最小二乘回归模型和人工神经网络模型。By taking advantages of the partial least square method for effective treatment of the multi-correlations between independent variables and BP neural network for effectively solving the nonlinear problems, a partial least square method and BP neural network coupled prediction model was developed. The application of the model to prediction of the pH value of urban acid rains shows that the precision of the model is high, and that it is superior to the partial least square regression model and the artificial neural network model.

关 键 词：酸雨 PH值 偏最小二乘回归 神经网络 预测模型 

分 类 号：X517[环境科学与工程—环境工程]