Miroshnyk V. Short-term forecasting of the electrical load of the power systems using deep learning neural network.

The оbject of research: total electrical load of the power system. The purpose of research: improvement of methods, development of models and tools to increase the accuracy and reliability of the results of short-term forecasting of the total electrical load of the power system with a significant share of energy-intensive enterprises and renewable energy sources. Methods of research: mathematical statistics, methods of numerical optimization, methods of time series analysis, machine learning methods, mathematical modeling, computer modeling. Theoretical and practical results and innovations: are in the developed architecture of deep learning neural network, the use of which increases the accuracy and stability of short-term forecasting of total electrical load; recommendations for taking into account the schedules of electricity consumption of energy-intensive enterprises in forecasting the total electrical load of the power system; recommendations for improving the quality of detection and replacement of anomalous data in the time series of electrical load; recommendations for improving the accuracy and stability of short-term forecasting of electricity supply by RES power plants using the developed deep learning neural network. A subject degree of introduction: Effectiveness of implantation: obtained a significant technical and economic effect, which is in increasing the reliability of the power system and reducing the cost of imbalances in the wholesale electricity market. Sphere of use: power systems.