基于偏最小二乘回归和浊度补偿的化学需氧量监测传感器的算法研究
发布时间：2019-01-02 08:29
【摘要】：随着水资源污染日益加剧,政府对水质监测的力度不断增强,一套具备快速、精确、可在线化监测功能的水质监测设备是提高监测水平和执法效率的关键所在.化学需氧量(COD)是反映水体中有机还原性污染物含量的重要指标,它是各国评价水体受污染程度的综合指标之一,所以水质COD监测传感器是一类非常重要的水质监测设备.目前,国内外的水质COD监测传感器种类繁多且性能各异,但检测水体COD的主要方法可大致分为两类:第一类是根据化学分析方法来检测COD;第二类是通过物理方法进行检测.化学方法主要以重铬酸钾法与高锰酸盐指数法为代表,其特点是测量精度高、耗时较长、对水样的预处理过程复杂、所用化学试剂易产生二次污染.物理方法主要以紫外吸收法(简称UV法)为代表,其特点是分析灵敏度高、不需要对水样进行预处理、无二次污染、操作简便、分析速度快.现如今各水体经常有突发性水污染事件发生,而且对水质的监测不能间断,所以用化学分析方法设计出的水质COD监测传感器已经不能满足上述要求了.同时,以紫外吸收法为原理所设计的水质COD监测传感器也还存在很多缺陷,所以本文的研究目的是改进以紫外吸收法为原理的光谱水质COD监测传感器的算法,进而使COD光谱监测传感器的测量精度提高、适用范围拓广,以便生产出准确、方便、可在线化连续监测的水质传感器.本文首先系统地介绍了紫外吸收法的基本原理,以及紫外吸收法中几种常用方法的发展历程和优缺点,并详细介绍了偏最小二乘回归的基本原理和相关性质.本文分析了前人的研究成果,在以前的研究中很多研究者是用待测液体在某一特定波长处的紫外吸光度来预测水样中的COD值:也有一部分研究者在此基础上考虑了浊度对特定波长处吸光度的影响,并对该特定波长的吸光度做了浊度补偿:还有一部分研究者是考虑用待测液体在多个指定波长处的吸光度来预测水样的COD值,但还没有研究者考虑过用待测液体在多个指定波长处的吸光度建模的同时又考虑浊度的影响.为了进一步提高COD监测的准确性和适用范围,本文提出了一种基于偏最小二乘回归和浊度补偿的COD检测算法.该算法是通过将全光谱吸光度检测法与偏最小二乘回归相结合来预测水体中的化学需氧量,同时考虑了浊度对建模所使用的吸光度(自变量)的影响,并对浊度的影响进行了补偿.通过实验分析表明:该方法对不同类型的污水均适用,检测的平均相对误差在5%以内,而且通过对比发现模型的预测精度明显优于未经浊度补偿的偏最小二乘回归模型,这为开发出一款适应性强且可在线化监测的水质COD传感器提供了算法依据.
[Abstract]:With the increasing pollution of water resources, the government is increasing the intensity of water quality monitoring. A set of water quality monitoring equipment with rapid, accurate and on-line monitoring function is the key to improve the monitoring level and law enforcement efficiency. Chemical oxygen demand (COD) is an important index to reflect the content of organic reducing pollutants in water. It is one of the comprehensive indexes to evaluate the degree of water pollution in various countries, so the water quality COD monitoring sensor is a kind of very important water quality monitoring equipment. At present, there are many kinds of water quality COD monitoring sensors at home and abroad, but the main methods of detecting water COD can be divided into two categories: the first is to detect COD; by chemical analysis method, and the second is to detect COD; by physical method. The chemical methods are mainly represented by potassium dichromate method and permanganate index method, which are characterized by high measurement accuracy, long time consuming, complex pretreatment process of water samples and easy secondary pollution of the chemical reagents used. The physical method is mainly represented by ultraviolet absorption method (UV), which is characterized by high analytical sensitivity, no pretreatment of water sample, no secondary pollution, simple operation and fast analysis speed. Nowadays, there are often sudden water pollution events in various water bodies, and the monitoring of water quality can not be interrupted. Therefore, the water quality COD monitoring sensor designed by chemical analysis method can not meet the above requirements. At the same time, the design of water quality COD monitoring sensor based on ultraviolet absorption method also has many defects, so the purpose of this paper is to improve the algorithm of spectrum water quality COD monitoring sensor based on ultraviolet absorption method. Thus, the measurement accuracy of the COD spectrum monitoring sensor is improved and the applicable range is extended to produce an accurate, convenient and on-line water quality sensor for continuous monitoring. In this paper, the basic principle of ultraviolet absorption method is introduced systematically, and the development history, advantages and disadvantages of several common methods in ultraviolet absorption method are introduced, and the basic principle and related properties of partial least square regression are introduced in detail. In this paper, the previous research results are analyzed. In previous studies, many researchers have used the UV absorbance of the liquid to be measured at a particular wavelength to predict the COD value in water samples: some researchers have also considered the effect of turbidity on absorbance at a particular wavelength. And the turbidity compensation for the absorbance of the specific wavelength is given: some researchers are considering using the absorbance of the liquid to be measured at several specified wavelengths to predict the COD value of the water sample. However, no researchers have considered the effect of turbidity on the absorbance modeling of the liquid to be tested at several specified wavelengths at the same time. In order to improve the accuracy and application range of COD monitoring, a COD detection algorithm based on partial least square regression and turbidity compensation is proposed in this paper. In this algorithm, the full spectral absorbance detection method and partial least square regression are combined to predict the chemical oxygen demand (COD) in water, and the influence of turbidity on the absorbance (independent variable) used in modeling is considered. The effect of turbidity is compensated. The experimental results show that the method is suitable for different types of sewage, and the average relative error is less than 5%, and the prediction accuracy of the model is obviously better than that of the partial least square regression model without turbidity compensation. This provides an algorithm basis for the development of an adaptive and on-line water quality COD sensor.
【学位授予单位】：四川师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212
本文编号：2398264
[Abstract]:With the increasing pollution of water resources, the government is increasing the intensity of water quality monitoring. A set of water quality monitoring equipment with rapid, accurate and on-line monitoring function is the key to improve the monitoring level and law enforcement efficiency. Chemical oxygen demand (COD) is an important index to reflect the content of organic reducing pollutants in water. It is one of the comprehensive indexes to evaluate the degree of water pollution in various countries, so the water quality COD monitoring sensor is a kind of very important water quality monitoring equipment. At present, there are many kinds of water quality COD monitoring sensors at home and abroad, but the main methods of detecting water COD can be divided into two categories: the first is to detect COD; by chemical analysis method, and the second is to detect COD; by physical method. The chemical methods are mainly represented by potassium dichromate method and permanganate index method, which are characterized by high measurement accuracy, long time consuming, complex pretreatment process of water samples and easy secondary pollution of the chemical reagents used. The physical method is mainly represented by ultraviolet absorption method (UV), which is characterized by high analytical sensitivity, no pretreatment of water sample, no secondary pollution, simple operation and fast analysis speed. Nowadays, there are often sudden water pollution events in various water bodies, and the monitoring of water quality can not be interrupted. Therefore, the water quality COD monitoring sensor designed by chemical analysis method can not meet the above requirements. At the same time, the design of water quality COD monitoring sensor based on ultraviolet absorption method also has many defects, so the purpose of this paper is to improve the algorithm of spectrum water quality COD monitoring sensor based on ultraviolet absorption method. Thus, the measurement accuracy of the COD spectrum monitoring sensor is improved and the applicable range is extended to produce an accurate, convenient and on-line water quality sensor for continuous monitoring. In this paper, the basic principle of ultraviolet absorption method is introduced systematically, and the development history, advantages and disadvantages of several common methods in ultraviolet absorption method are introduced, and the basic principle and related properties of partial least square regression are introduced in detail. In this paper, the previous research results are analyzed. In previous studies, many researchers have used the UV absorbance of the liquid to be measured at a particular wavelength to predict the COD value in water samples: some researchers have also considered the effect of turbidity on absorbance at a particular wavelength. And the turbidity compensation for the absorbance of the specific wavelength is given: some researchers are considering using the absorbance of the liquid to be measured at several specified wavelengths to predict the COD value of the water sample. However, no researchers have considered the effect of turbidity on the absorbance modeling of the liquid to be tested at several specified wavelengths at the same time. In order to improve the accuracy and application range of COD monitoring, a COD detection algorithm based on partial least square regression and turbidity compensation is proposed in this paper. In this algorithm, the full spectral absorbance detection method and partial least square regression are combined to predict the chemical oxygen demand (COD) in water, and the influence of turbidity on the absorbance (independent variable) used in modeling is considered. The effect of turbidity is compensated. The experimental results show that the method is suitable for different types of sewage, and the average relative error is less than 5%, and the prediction accuracy of the model is obviously better than that of the partial least square regression model without turbidity compensation. This provides an algorithm basis for the development of an adaptive and on-line water quality COD sensor.
【学位授予单位】：四川师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP212
【参考文献】
相关期刊论文 前10条
1 赵立军;;COD在线测量中温度影响的消除方法研究[J];仪表技术与传感器;2015年12期
2 罗继阳;魏彪;汤斌;赵敬晓;吴德操;米德伶;俞鹏炜;;水质COD检测用光谱法的水体环境浊度解算方法[J];环境科学与技术;2015年S2期
3 毕新胜;贾金亮;王玉刚;李玉强;曾小辉;王斌;李长玲;;近红外光谱技术在籽棉回潮率检测系统的应用[J];自动化仪表;2015年09期
4 张颖萍;董前民;张章源;滕斌;泮进明;;双波长紫外光谱法检测猪场污水实验研究[J];中国农机化学报;2015年03期
5 陈岩;闫云浩;谭婷;郭宏;;水质监测多参数的标定研究[J];仪表技术与传感器;2014年06期
6 侯迪波;张坚;陈泠;黄平捷;张光新;;基于紫外-可见光光谱的水质分析方法研究进展与应用[J];光谱学与光谱分析;2013年07期
7 赵友全;李玉春;郭翼;顾柏军;杨震;;基于光谱分析的紫外水质检测技术[J];光谱学与光谱分析;2012年05期
8 王玉峰;陈克复;莫立焕;周志明;;造纸废水化学需氧量的双波长紫外光谱法测定[J];华南理工大学学报(自然科学版);2012年02期
9 吴国庆;毕卫红;吕佳明;付广伟;;近红外透射和紫外吸光度法检测水质化学需氧量的研究[J];光谱学与光谱分析;2011年06期
10 夏阿林;叶华俊;周新奇;王健;;基于粒子群算法的波长选择方法用于苹果酸度的近红外光谱分析[J];分析试验室;2010年09期
相关硕士学位论文 前2条
1 张国强;UV法测量水质COD技术研究[D];电子科技大学;2007年
2 余建平;基于改进神经网络的污水处理参数软测量模型研究[D];重庆大学;2005年
，本文编号：2398264
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2398264.html
