ANALYSIS OF ELECTRICITY CONSUMPTION USING THE COMPONENT METHOD OF PERIODICALLY CORRELATED RANDOM PROCESSES

Abstract

Contemporary energy systems, considering the diverse challenges emerging within energy infrastructure, require advanced analytical methodologies for electricity consumption forecasting. Traditional statistical approaches prove insufficient for modeling dynamic multi-scale temporal structures of electricity consumption signals aimed at predicting electrical loads in residential households.

This research presents a comprehensive approach to electricity consumption analysis utilizing the mathematical framework of periodically correlated random processes (PCRP), specifically employing the component method. The mathematical foundation of the methodology consists in representing electricity consumption signals as PCRP models with decomposition into constituent elements: deterministic trend components, periodic components of cyclical variations, and stochastic components of random deviations. Component analysis enables the identification of latent consumption patterns through decomposition of periodic characteristics. Therefore, the proposed method allows for the elimination of limitations inherent in traditional stationary models.

Empirical validation was conducted using a comprehensive dataset of residential electricity consumption spanning the period from July to August 2025. Experimental data demonstrated pronounced repetitive characteristics with systematic daily periodicity, confirming the theoretical premise regarding daily component dominance. Three-dimensional visualization of results revealed complex interaction dynamics between different frequency components of electrical load signals. Spectral analysis exhibited characteristic distribution with maxima for low-frequency components corresponding to daily harmonics.

The obtained results can be utilized for residential electrical load forecasting and enable both short-term and medium-term energy consumption predictions. This is significant not only for forecasting residential electrical loads, but also for optimizing electrical energy resources and managing intelligent networks.