基于主成分分析-BP神经网络法的松花江哈尔滨段水质评价研究

发布时间：2018-05-21 15:49

  本文选题：水质评价 + BP神经网络　； 参考：《哈尔滨师范大学》2015年硕士论文

【摘要】：在社会高度发达的今天,经济发展的速度越来越快,人类无止境的向自然进行索取,肆意的破坏着我们的家园,水环境的污染首当其冲。水环境质量的改善刻不容缓。对水质进行评价可以得知水环境质量的现状,为水环境的治理提供科学的依据。作为全国七大水系之一的松花江,承载着周围的生活和生产。本研究以松花江哈尔滨段为研究对象,针对松花江哈尔滨段的具体情况,在综合考虑了影响松花江水质的各种自然及人文因素的基础上,结合重要性原则本文选取了包括化学需氧量、五日生化需氧量以及氨氮在内的18个水质监测指标作为松花江水质评价的基础指标。通过主成分分析法对研究指标进行筛选,最终选取了化学需氧量、五日生化需氧量、高锰酸盐指数、氨氮、总磷和石油类作为BP神经网络评价水质的水质指标。研究中采用的水质数据为2009-2013年朱顺屯、阿什河口内、阿什河口下、呼兰河口内、呼兰河口下和大顶子山6个断面指标的月度监测值。分为按年综合评价和按枯水期、平水期和丰水期时期来进行评价。采用的水质评价方法为主成分分析法和BP神经网络的方法,通过主成分分析法对各断面的主成分值进行排名,得出各断面的污染先后顺序,其次选取污染指标中的主成分,为BP神经网络模拟水质提供指标依据。通过训练BP神经网络,得出最优的评价模型,通过模型分别对各个时期的断面进行评价,对各断面的污染等级进行划分,得出各断面的水质分类等级。本文主成分分析法选用的软件为SPSS22.0,BP神经网络法选用的是MATLAB2010a。评价结果表明:六个断面中水质最差的为阿什河口内断面,五年内排名均在最后一名,水质均为Ⅴ类水质,水质最好的为朱顺屯断面,为Ⅱ-Ⅲ类水质,其余四个断面水质均较好,基本达到规划的水质要求。通过两种方法来评价松花江哈尔滨段的水质情况,为改善和治理水质提供科学依据。
[Abstract]:In the highly developed society today, the speed of economic development is getting faster and faster. Human beings are taking endless demands from nature, wantonly destroying our homes, and the pollution of water environment bears the brunt. It is urgent to improve the quality of water environment. To evaluate the water quality, we can know the present situation of the water environment quality, and provide scientific basis for the water environment management. As one of the seven major water systems in the country, the Songhua River carries the life and production around it. This study takes Harbin section of Songhua River as the research object, according to the concrete situation of Harbin section of Songhua River, on the basis of synthetically considering all kinds of natural and human factors that affect the water quality of Songhua River. According to the principle of importance, 18 water quality monitoring indexes, including chemical oxygen demand, five days biochemical oxygen demand and ammonia nitrogen, were selected as the basic indexes for water quality evaluation of Songhua River. The chemical oxygen demand, five days biochemical oxygen demand, permanganate index, ammonia nitrogen, total phosphorus and petroleum were selected as the water quality indexes of BP neural network. The water quality data used in the study are monthly monitoring values of 6 cross section indexes in Zhushuntun, Ashi Estuary, Hulan River Estuary, Hulan River Estuary and Dadingzi Mountain from 2009 to 2013. It can be divided into annual comprehensive evaluation and dry, plain and abundant water periods. The methods of water quality evaluation are principal component analysis and BP neural network. The principal component value of each section is ranked by principal component analysis, the pollution sequence of each section is obtained, and the principal component of pollution index is selected. It provides index basis for BP neural network to simulate water quality. Through training BP neural network, the optimal evaluation model is obtained. The section of each period is evaluated by the model, the pollution grade of each section is divided, and the water quality classification grade of each section is obtained. In this paper, the software of principal component analysis is SPSS 22.0 BP neural network method and MATLAB2010a. The results show that the worst water quality of the six sections is the Ashe estuary section, which ranks last in five years. The water quality is category V, the best water quality is the Zhushuntun section, and the water quality is class 鈪，

本文编号：1919825