The system imbalances are becoming severer due to increasingly integration of renewable energy sources. Lack of short-term predictions is making balancing the system more difficult for the system operator to handle and increases costs. In this context, we have developed an algorithm for short-term forecasting of system imbalances, with the aim of improving economic efficiency, thereby reducing the cost of balancing and optimal use of resources.
Within the framework, the technical background of balancing the system, which are carried out by operators with the help of various mechanisms, was investigated. Based on the researched area, we gained a deeper understanding of the key problems and challenges related to the balancing of the power system. We explored the theoretical foundations of forecasting and described the most affective approaches for the problem under consideration.
During the work, historical data was analysed and processed, which served as the basis for the development and learning of a short-term forecast model based on machine learning methods. Due to the lack of short-term forecasts, we developed a correction model based on simple mathematical principles. this allowed us to directly compare the accuracy of the developed models.
The analysis of the results shows that, at the current stage of development, a simple correction model is more accurate than a machine learning model. The reason for the slightly lower accuracy of the machine learning model is the limited set of input parameters. By including additional input parameters, especially the weather forecast and the forecast of production from renewable energy sources, we expect an improvement in the accuracy of the machine learning model. The implementation of the developed algorithm will enable operators to handle system imbalances with enhanced operational efficiency.
|
