Data Driven Update of Load Forecasts in Smart Power Systems using Fuzzy Fusion of Learning GPs

Alamaniotis, Miltiadis; Martinez-Molina, Antonio; Karagiannis, Georgios

Abstract

One of the pillars in developing smart power systems is the use of load forecasting methods. In particular load forecasting accommodates decision making pertained to the operation of power market. In this paper, a new method for real-time updating very short-term load forecasting is proposed. The goal of the method is to accurately predict the load demand value in the next 5 minutes and accordingly update the daily forecast. To that end, the proposed method implements an ensemble of homogeneous learning Gaussian processes which are trained on slightly different training datasets. The predicted values are then fused using a fuzzy inference system in order to obtain a single value which is used to correct the precomputed forecast. The proposed method is tested on a set of real-world data taken from a major US area and is benchmarked against the naïve forecasting method. Results highlight the superiority of our method against the benchmarked method exhibiting an increase in forecasted accuracy by 50% in most cases.

Citation

Alamaniotis, M., Martinez-Molina, A., & Karagiannis, G. (2023, June). Data Driven Update of Load Forecasts in Smart Power Systems using Fuzzy Fusion of Learning GPs. Presented at 2021 IEEE Madrid PowerTech, Madrid, Spain

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2021 IEEE Madrid PowerTech
• Start Date: Jun 28, 2023
• End Date: Jul 2, 2021
• Acceptance Date: Jul 2, 2021
• Online Publication Date: Jul 29, 2021
• Publication Date: 2021
• Deposit Date: Aug 12, 2021
• Publicly Available Date: Aug 12, 2021
• DOI: https://doi.org/10.1109/powertech46648.2021.9494757
• Public URL: https://durham-repository.worktribe.com/output/1139114