Application of machine learning models to predict energy consumption in smart home systems

V.O. Haidukevych, A.Yu. Doroshenko

Abstract


The article investigates the application of machine learning methods to forecast energy consumption in the con text of smart home systems. The research is based on the internationally renowned PSML (Power System Ma chine Learning) time series dataset, which includes information on electricity consumption, generation, balanc ing and load forecasting in the context of a decarbonized energy network. The PSML dataset is characterized by high detail and covers different time scales - from hourly to daily values, which allows assessing both short-term and medium-term trends in energy consumption. The paper provides a comparative analysis of classical and modern machine learning methods, including regres sion, classification and clustering, for load forecasting and identifying patterns in electricity consumption in the domestic environment. Particular attention is paid to optimizing models for working with big data, processing gaps and anomalies, as well as integrating forecasts into automatic smart home energy management systems.

Prombles in programming 2025; 3: 29-38


Keywords


machine learning; forecasting; smart home; energy optimization; PSML

References


Hyndman, R. J., & Athanasopoulos, G. (2018). Forecasting: Principles and practice (2nd ed.).

Weron, R. (2014). Electricity price forecasting: A review of the state-of-the-art with a look into the future. International Journal of Forecasting, 30(4), 1030–1081.

Ryu, S., Noh, J., & Kim, H. (2017). Deep neural network based demand side short term load forecasting. Energies, 10(1).

Open source PSML dataset

Sinitsyn, I. P., Shevchenko, V. L., Doroshenko, A. Yu., & Yatsenko, O. A. (2025). Research of software solutions for forecasting electricity production and consumption in Ukraine based on machine learning methods. Problems of programming and information technologies, 15(1), 45–58.

Vyshnevskyy, O. V., & Zhuravchak, L. (2023). Machine Learning Methods to Increase the Energy Efficiency of Buildings. Computer Systems and Networks, 14, 189–209.

Li, W., & Wang, J. (2019). Electricity consumption forecasting using a hybrid model based on data preprocessing and long short term memory neural network. Energy, 169, 1099–1110.

Ahmed, R., & Khalid, M. (2020). A review on the selected applications of forecasting models in renewable energy systems. Renewable and Sustainable Energy Reviews, 124, 109792.

Khac, T. N., & Bui, T. V. (2021). Load forecasting using machine learning: A comparative study of neural networks and random forests. Energy Reports, 7, 394–403.


Refbacks

  • There are currently no refbacks.