An Improvement of Power Demand Prediction Method using Weather Information and Machine Learning: A Case of a Clinic in Japan (II)

• Published in: 2023 11th International Conference on Smart Grid (icSmartGrid) pp. 1 - 6
• Main Authors: Inagata, Tomoya, Matsunaga, Keita, Mizuno, Yuji, Kurokawa, Fujio, Tanaka, Masaharu, Matsui, Nobumasa
• Format: Conference Proceeding
• Language: English
• Published: IEEE 04.06.2023
• Subjects: clinic; Demand response; load prediction; machine learning; Meteorology; Power generation; Power systems; Predictive models; Renewable energy sources; Research and development; Smart grids; Supply and demand; Temperature distribution; weather information
• DOI: 10.1109/icSmartGrid58556.2023.10170940

In Japan, power generation from renewable energy sources has been promoted since the Great East Japan Earthquake in 2011. For this reason, the installing of renewable energy is increasing in Japan. However, the amount of power generated by renewable energies is influenced depending on natural conditions. To ensure a stable supply of electricity, it is important to keep a balance between supply and demand. Therefore, research and development of a demand response is becoming increasingly important. Hospitals and clinics, which are among the most energy-consuming types of medical facilities using renewable energy systems, need to predict an electricity demand to consider carrying out a demand response. This paper proposes a method to improve the accuracy of an electricity demand prediction for a clinic. A neural network is used as a prediction method, and the predictors consist of day of the week and temperature data by Japan Meteorological Agency. As a result, it is clarified that the proposed method is close to \sqrt{\pi/2}\approx 1.253 , which is the value to be evaluated when the error is normally distributed.