MACHINE LEARNING APPROACH FOR ENERGY CONSUMPTION ESTIMATION IN DISTRICT HEATING SYSTEMS

Abstract

This paper explores the application of various machine learning methods for time series estimation in district heating systems. The focus is on predicting heat load using supervised machine learning techniques, such as artificial neural networks and the Random Forest algorithm. Input parameters are derived from the SCADA system, including outdoor air temperature, water temperature, and pressure in the primary and secondary circuits.The development of advanced predictive models enables a better analysis of energy consumption patterns, which is crucial for improving the efficiency of district heating systems. The paper examines different machine learning approaches to identify the models that best meet the requirements for prediction in real operating conditions.The expected contribution of this study lies in establishing a foundation for further automation and optimization of district heating systems, potentially leading to reduced operational costs and energy consumption, as well as enhancing the sustainability of energy systems.