Research On Abnormal Event Detection Of Urban Water Supply Network Based On Pipe Pressure Data

In recent years,with the rapid growth of China’s population and the development of social economy,the scale of urban water supply is expanding,and the structure of urban water supply pipe network is becoming more and more complex.At present,the structure of urban water supply network is obsolete,and the planning is unreasonable.The aging corrosion of buried pipelines is serious,the water supply pipe network burst frequently,which causes a large amount of water loss,seriously impact on social and economic development and residents’ lives.How to effectively detect the abnormal events in the urban water supply pipe network system and to locate them quickly and accurately has become a hot topic in the research of domestic and foreign water supply enterprises and academia.In this thesis,the abnormal events in the urban water supply pipe network system are detected and located.The specific work is as follows:Firstly,based on the historical monitoring Data of the Supervisor Control and Data Acquisition(SCADA)system,the fluctuation characteristics of pressure parameters at each monitoring point are analyzed,and the normal distribution probability function is introduced to define parameter threshold and determine abnormal indexes.Based on the actual pipe burst accident records of a large independent Metering Area(DMA)in S City,the detection performance of each monitoring point in the pipe network system was analyzed.The experimental results show that the anomaly detection method based on threshold value can effectively detect large pipe burst accidents in the pipe network,but it is susceptible to the background noise in the environment of pipe network pressure fluctuation.Secondly,in view of the defects of traditional point anomaly detection methods that are sensitive to parameters,low precision and vulnerable to interference,this paper proposes a DMA partition leakage detection method based on soft voting.VMD-SE denoising method was used to decompose and reconstruct the pressure signal,and the background noise in the historical pressure data was eliminated.The sudden noise was super-placed to the normal pressure data to simulate the leakage accident,and then the leakage data set was obtained.On this basis,combined with the feature extraction and soft voting classification mechanism,a high precision leakage identification model was established.The experimental results show that the proposed method has lower false alarm rate and higher accuracy.Finally,in order to locate the leakage points of pipe network as soon as quickly,a regional leakage location method based on improved K-means clustering was studied.Space of k-means clustering algorithm considering monitoring spatial and non-spatial attribute,the introduction of information entropy to calculate the clustering attribute weights,on the clustering and classification of water supply network monitoring leakage area,finally the regional monitoring using center of gravity iteration model to solve the leakage point coordinates,realize regional leakage accurate positioning.The results of an example show that the method performs well in the location of leakage loss and has high positioning accuracy.