From a forecasting methodology for the electric energy consumption of mono-towns to its sustainability

Abstract

The need of Russia for a resource-innovative energy strategy for the period up to 2035 makes it urgent to create a methodology of strategic forecasting for the electric energy consumption of mono-towns, which will result in the efficient development of a power supply system, ensuring territories’ sustainability. For this purpose, it is necessary to consider a mono-town as a complete system, a living organism with a definite life cycle and structure of energy consumers. The authors offer a technocoenosis approach to carry out the ranking and structural analysis of the electric energy consumption of one of the Sverdlovsk region’s mono-towns, taking into account dynamics in population, enterprises, organizations and institutions for the period of 5 years. The provided analysis makes it possible to judge the technocoenosis optimality using characteristic exponent β, depending on the structural features of the territory. The authors developed an algorithm for electric energy consumption forecasting, based on Support Vector Machines (SVM), which takes into account the electric energy of mono-towns’ consumers and the climatic factors. Forecasting accuracy was achieved using cross-validation of the input data in order to optimize the training model. The corresponding changes in electric energy consumption, when a characteristic exponent is optimal, will result in a target forecast that provides the sustainable development of mono-towns.