COVID-19 Salgınının Gün Öncesi Piyasası Elektrik Talep Tahminine Etkisinin İncelenmesi


AJDER A. (Executive)

Project Supported by Higher Education Institutions, 2021 - 2022

  • Project Type: Project Supported by Higher Education Institutions
  • Begin Date: August 2021
  • End Date: August 2022

Project Abstract

The COVID-19 has effected all over the world since the first months of 2020. Leaving aside its health-related parts, the impact of the pandemic on each sector has been different; these effects have to be evaluated by experts in their subjects. From the perspective of the energy sector, the closures taken by the decision makers during the COVID-19 period caused huge changes in the electricity demand profiles of the countries. These changes concern both electricity grid operators and system participants. As it is known, forecasting in the day ahead market is crucial for managing the uncertainty in intraday and balancing market. To make error in these forecasting will expose system operators and participants to technical and financial risks, causing the economic effects of the epidemic to be felt even more heavily for the energy sector.

In this project, Artificial Neural Network (ANN) models, which have proven to be an effective solution in nonlinear problems, are used to analyze the impact of COVID-19 on electricity demand. The performances of Recurrent Neural Networks (RNN) based Long Short Term Memory (LSTM) and Gated Repetitive Unit (GRU) architectures, which are preferred among deep learning algorithms, especially in the analysis of time series, are compared. In comparison, of the best architectures of all models, LSTM gives 2.07% better results for the validation dataset, while the computational efficiency of the GRU, which is less complex structure than LSTM, is better.