| From the point of the total water resources, our country ranks sixth in the world. Although the ranking of total water resources is near the top, but the population of our country is large, so China is a water-scarce county, per capita water occupancy volume is much lower than the world average. Meanwhile, the sustainable development of China’s industrialization has produced a lot of pollution. Water pollution has caused significant damage to ecological environment and human health are also subject to a greater threat. In addition, water pollution makes the problem of water shortage more serious. If we can provide early warning information effectively and identify the contaminant before the expansion of water pollution accident, it will help to make rapid response measures and protect water resources. Therefore, it is an urgent problem to establish a system of water quality anomaly detection and contaminants classification.The monitoring of water quality is usually conducted by using off-line or on-line water quality sensors. There are many kinds of water quality sensors used in water quality monitoring. Generally speaking, it can be classified into two types. One type is to detect specific pollutants or a class of pollutants. Although the sensor for detecting specific pollutants can quantitatively determine the specific pollutants in a low concentration, but there are some shortcomings, such as poor anti-interference, high price, short life and so on. For these reasons, such sensors are generally difficult to widely use. The other type of sensor is different from the sensor for detecting specific pollutants, which is not for detection of specific or a class of contaminant, but reflects the comprehensive condition of water quality. Such as pH, conductivity, temperature and other water quality sensor, we call these conventional water quality sensors.There are many kinds of contaminants in the polluted water. The existing testing instruments are difficult to detect each contaminant. Instruments under development used to detect specific types of contaminants are expensive. The sensor used for detection of conventional water quality parameters are low expense, mature detecting technique and stable performance. There is a certain relationship between conventional water quality parameters and the specific kinds of contaminants. The ability to characterize contaminant can be further improved by data analysis and data mining techniques. These show the advantages of the conventional water quality sensors and the feasibility of the conventional water quality parameters used for water quality detection and contaminants classification. Conventional water quality sensor has a series of advantages for water quality monitoring, which makes the application of conventional water quality sensors in water quality monitoring more and more widely.In consideration of the importance of water quality anomaly detection and the advantages of conventional water quality sensor in water quality monitoring. This paper studies on water quality detection by using conventional water quality sensors. In this paper, the influence of the background water quality parameter’s fluctuation is eliminated by the data standardization. The polygon area method is used to fuse the information of multiple water quality indexes to improve the performance of anomaly detection. In order to judge the abnormal water quality, we set the threshold of polygon area based on X-control chart principle and compare the polygon area calculated by water samples’ quality parameters to the threshold of polygon area. Heavy metals, phenols, anilines and pesticides three representative contaminants were selected in the experiments of water quality anomaly detection and obtained good results.In order to provide decision support for emergency measures’making in water pollution accident, it is necessary to identify the type of contaminants as soon as possible after water quality anomaly detection. The conventional methods for contaminants identification is based on laboratory analysis. Although it can be qualitative and quantitative analysis of contaminants accurately, but it is time-consuming in the actual detection.There are some drawbacks in rapid response. Online water quality sensor has the advantages of real-time online, simple and fast measurement of water quality parameters. The same water quality sensor’s response to a type of contaminants is similar, and there are differences in response among different types of contaminants. There is a certain correlation between the water quality parameters and the types of contaminants. Therefore, the technology of data mining can be used to analyze the data of conventional water quality parameters to realize the classification of contaminants. This paper has established a system of water contaminants classification and identification based on a large amount of water quality monitoring data.The six water quality parameters which are selected in this paper are fast and easy to be detected, which can represent the characteristics of water quality. The detection technology of these six water quality parameters is mature. The used instruments are simple which can achieve integration. If combined with network technology, real-time and on-line monitoring of water quality can be realized in different water bodies and locations, and provide timely and effective information forecasting and decision support for the prevention and treatment of water accidents.The dissertation includes three chapters:Chapter 1 IntroductionDo a general introduction for the status of water resources and water pollution situation of China. This chapter introduces the domestic and foreign research status on the water quality anomaly detection, the method of water quality parameter detection, the detection algorithm of water quality anomaly based on the related literature. This chapter introduces and summarizes the method for water quality anomaly detection and contaminants classification, illustrates the theoretical basis, the purpose and innovation of this research in detail.Chapter 2 Study on water quality anomaly detection based on the six conventional simple water quality parametersWater quality anomaly detection is the key to water quality monitoring and early warning system. It is very important to detect water pollution timely and accurately, to provide early warning, to improve the treatment ability of sudden water pollution accident, to prevent water pollution and protect water resources. Through the study on the relationship between the six conventional water parameters (temperature, pH, dissolved oxygen, conductivity, ORP, UV254) and contaminants. It is found that the six water quality parameters have varying degrees of response to various contaminants, which provides a basis for the study of water quality anomaly detection based on the conventional water quality parameters.This chapter establishes a method of water quality detection by using six conventional water quality parameters. The influence of the background water quality parameter’s fluctuation is eliminated by the data standardization. The polygon area method is used to fuse the information of multiple water quality indexes to improve the performance of anomaly detection. In order to judge the abnormal water quality, we set the threshold of polygon area based on X-control chart principle and compare the polygon area calculated by water samples’quality parameters to the threshold of polygon area. Heavy metals, phenols, anilines and pesticides three representative contaminants were selected in the experiments of water quality anomaly detection and obtained good results.Chapter 3 Study on contaminants classification based on the six conventional simple water quality parametersIn this chapter, contaminants classification is based on the measurement of water quality parameters and cluster analysis of data mining. The cosine distance is used as the similarity measure to classify the types of contaminants. This chapter takes data standardization and optimizes the combination of water quality parameters. The difference of cosine distance between some contaminants is small. Cosine discrimination was posed as similarity measure for contaminants classification. Finally, phenols and anilines, heavy metals and pesticides are used in the experiment of contaminants classification. This method exhibits excellent classification performance. |
PDF Full Text Request