Analysis of Theft of Housing and Communal Services Resources Based on Neural Network Technologies

This article discusses the main ways of manipulating intelligent metering devices for housing and communal services. The advantages and disadvantages of intelligent metering devices are considered. The analysis of articles containing modern algorithms and methods for combating theft of electricity is given. Articles on improving the reliability of intelligent metering devices based on neural network analysis are also considered. A proprietary algorithm for detecting theft of housing and communal services resources has been developed. The algorithm is based on an individual consumer model, the main feature of which is the mutual consumption of housing and communal services resources and consideration of internal and external factors. An experiment was also conducted to detect the theft of electricity. Based on real data, a sample of 800 consumption cases was modeled, 400 of which are normal consumption and 400 are theft of electricity. The theft of electricity was simulated by underestimating the meter readings. Then, with the help of a direct distribution neural network, clustering was done, cases of normal consumption were separated from theft. The optimal structure of the neural network was also determined.