Abnormal Data Analysis and Alarm Strategy for Urban Water Supply Network Research

The urban water supply network is one of the essential infrastructures and an important part of the urban water supply system. In view of the problems of false alarm and leakage in the existing urban water supply network monitoring and management system, an abnormal data detection method based on pressure and flow data fusion for urban water supply network is studied. A back propagation (BP) neural network model with good adaptivity is constructed using the flow and pressure history data to achieve high accuracy classification of normal, burst, underpressure and negative pressure data of water supply network with 89.98% recognition accuracy. Meanwhile, the abnormal data classifier developed based on the BP neural network model makes abnormal data alarm decision, classifies abnormal conditions into levels and sends alarm signals in time to reduce the false alarm leakage rate.