Forecasting volumes of electricity sales using a multilayer neural network

Abstract

Energy consumption is related to all types of human economic activity. Energy production significantly affects the state of the environment; therefore, high quality planning of electricity consumption is a priority issue in the development strategy of any country. All responsibilities for planning electricity consumption and purchasing electricity fall on energy supply companies. High-quality planning of electricity consumption will lead to a reduction in the costs of energy supply companies, which will have an impact on reducing the costs of creating final products for the average consumer. This article is devoted to the problem of increasing the efficiency of energy supply companies to meet the needs of household consumers. The work substantiates the necessity of planning electricity supplies in the necessary volume to ensure the rational use of energy and economic resources. The problem of performing forecasting by a person is described, when the result depends both on the analyst's qualification in the subject field and on his qualification in analysis methods. Arguments are presented regarding the use of modern methods of artificial intelligence - neural networks, which have the ability to learn on a set of data and allow to reveal hidden relationships and regularities between data. The article provides an analysis of the latest research and publications on the use of artificial neural networks in energy forecasting, which confirmed the feasibility of using the chosen method for a given problem. The process of using a multilayer neural network for forecasting the volume of electricity sales to consumers is described. The nature of the data describing the volume of monthly hourly electricity purchases for household consumers is given, an example of a time series is shown. The peculiarities of the input parameters for the neural network are determined: month of the year, day of the week, type of day (holiday or working), the average maximum value of hourly electricity purchase