基于人工神经网络样本优化法识别的河流溶解氧模型  

A model to evaluate do of river based on artificial neural network and stylebook

作　　者：陈丽华[1] 马生全[2] 李丽[3] 

机构地区：[1]西北民族大学实验中心,甘肃兰州730030 [2]海南师范大学数学与统计学院,海南海口571158 [3]西北民族大学化工学院,甘肃兰州730030

出　　处：《海南师范大学学报（自然科学版）》2008年第4期372-376,共5页Journal of Hainan Normal University（Natural Science）

基　　金：国家教育部"春晖"计划(S2004-1-62003)

摘　　要：以影响河流溶解氧的3个重要因素(流量、水温、pH)建立了基于L-M算法的BP神经网络模型,用此模型可成功预测黄河水中溶解氧(DO)的浓度.过程中选取四种方法对学习样本进行预处理,利用处理后的数据训练网络,结果表明采用方法四归一化后的数据预测DO含量效果最好,同时表明线性插值后学习样本的大小对网络预测的精密度有较大提高.Three important factors： the outflow, the water temperature and the pH, which influenced the DO （dissolved oxigen） of river directly were used to set up the BP neural network based on L-M algorithm. And the DO concentration of the Yellow River was predicted successfully with this neural network. Four methods were used to unitary the learning samples, and then the network was trained by them. The results showed that the network can predict the DO concentration satisfactorily which was unified by the fourth method, and the numbers of network＇s learning sample effected the precision of the network＇s prediction after many numbers were inserted in these learning samples lineally.

关 键 词：人工神经网络 Levenberg-Marpuardt算法 溶解氧(DO) 归一化 线性插值 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] X323[自动化与计算机技术—控制科学与工程]