基于T-S模糊神经网络的颍河水质时空变化特征分析  

Analysis of Characteristics of Ying River Water Quality Change of Time and Space Based on T-S Fuzzy Neural Network

作　　者：张文[1] 王莉[1,2] 

机构地区：[1]郑州大学水利与环境学院,河南郑州450001 [2]西安理工大学陕西省西北旱区生态水利工程重点实验室,陕西西安710048

出　　处：《环境科学与技术》2015年第12期254-261,共8页Environmental Science & Technology

基　　金：水体污染控制与治理科技重大专项;淮河流域(河南段)水生态修复关键技术研究与示范课题(2012ZX07204-004)

摘　　要：基于T-S模糊神经网络分析颍河(河南段)2007-2010年5个监测断面的水质指标监测数据,克服了过去仅用各级评价标准作为训练样本,导致训练样本数过少和不能构建检测样本的缺点,对实测数据仿真有很好的效果。应用主成分分析法,选出溶解氧、高锰酸盐指数、生化需氧量、氨氮、化学需氧量、总磷、六价铬、阴离子表面活性剂8项对水质具有重要影响的指标,建立适用于颍河的T-S模糊神经网络水质评价模型,对颍河水质时空变化特征进行分析。结果显示:颍河水质在近几年呈改善趋势;白沙水库断面水质最优,且稳定;周口康店断面水质较差,西华址坊在2007年第2季度水质恶化严重;沈丘纸店断面水质逐渐改善。表明了T-S模糊神经网络对水质时空变化特征分析的效果显著,也反映了颍河水质状况,为颍河水质监测、管理与控制提供依据。On the basis of T-S fuzzy neural network,five indexes of water quality monitoring data of Henan Section of Ying River from 2007 to 2010 were analyzed,which is helpful to the simulation of the measured data,and better than previous method that uses evaluation criteria at all levels,resulting in too little training samples and impossibility of building testing examples. Principal component analysis has been applied with 8 index badly influencing water quality,like DO,CODMn,BOD_5,NH_3-N,COD,TP,Cr（Ⅵ）,anionic surfactant to participate in the Ying River water quality assessment.Results indicate that the water quality of Ying River water improved continuously in recent years,with the best and more stable water quality in Baisha Reservoir Section,poor water quality in Zhoukoukangdian Section,and water quality of Xihuazhifang deteriorated seriously in the second quarter of 2007, while quality of Shenqiuzhidian Section gradually improved. It is proved that excellent effects on temporal and spatial analysis of water quality with T-S fuzzy neural network are obtained, which can reflect water quality of Ying River basin, and offer supports for water quality monitoring,management and control.

关 键 词：T-S模糊神经网络 颍河 水质评价 水质时空变化特征 主成分分析 

分 类 号：X824[环境科学与工程—环境工程]