The development of the information system for anomality detection in the utility meters data using self-organized maps

Abstract

In this article, a project has been developed for the modernization of the data analysis technology of the system for accounting for the consumption of utility resources. The need to improve the system is due to insufficient efficiency in identifying the facts of unaccounted consumption of utility resources. Automation of data analysis processes will be based on the development of an artificial neural network. A feedforward network based on a multilayer perceptron and consisting of 1 hidden layer was chosen as a model. The backpropagation algorithm was chosen as the method for training the neural network