Research Of Water Quality Outlier Detection Method Based On Local Minima Density

Water resource is an important resource for the sustainable development of human beings, and the water ecology is closely related to the human society. In recent years, water pollution accidents occur frequently in the Three Gorges Reservoir area, which brings great threat to people's lives and ecological environment. The health of the water environment of the Three Gorges Reservoir is related to national economic stability and development, at the same time, it is also the key of national security and stability. It has a very important strategic position.The main work of this thesis is based on the characteristics of clustering and local density of the outlier, and proposes an outlier detection algorithm based on the minimum density. Through the experimental analysis, this thesis verifies the effectiveness of the research method, and applies it on the Three Gorges Online Monitoring System, the main work and innovation of this thesis is as follow:1. The traditional methods based on nearest neighbor outlier detection have some weaknesses:performance is sensitive to parameter, and interpretability is not strong. In this thesis, based on the novel idea that outlier objects have lower density than their neighbors and relatively large distance from objects with higher density, this thesis propose a new algorithm for outlier detection to overcome the weakness above. This thesis compared the proposed method with other existing methods based on various types of synthetic datasets.2. Based on the proposed algorithm, this thesis considers the characteristics of water quality and focus on the real water quality data processing. It Includes data loss processing based on SVR regression, data standardization and PCA preprocessing, as well as the application of water anomaly detection. Parameter optimization metheod is aslo proposed. The experimental results show that the proposed algorithm can detect the different levels of water quality pollution effectively.3. The data-driven research on water quality anomaly detection is the basic work of water pollution control planning and integrated control in the Three Gorges Reservoir area. In order to reduce the the Three Gorges Reservoir water environment problem, in this thesis, we use the proposed algorithm to analyze the data water quality in the Three Gorges Reservoir area and apply it on the Three Gorges Online Monitoring System. We have used information technology to make efforts to sense the ecological environment of the Three Gorges.In this thesis, the research of water quality detection will be helpful to realize the real-time analysis of abnormal online detection system and lay the foundation for the informatization of water quality warning system.