Travail de fin d'études et stage[BR]- Travail de fin d'études : Development of machine learning-based surrogate model in the European power system[BR]- Stage

Abstract

This work centers on the development of a machine learning-based surrogate model to enhance the efficiency of simulating the European power system. Initially, a series of simulations will be conducted using Dispa-SET, a short-term dispatch optimization model employed to manage the operation of the European power grid. Dispa-SET is essential for balancing electricity supply and demand, ensuring grid efficiency, and minimizing operational costs by optimizing the dispatch of generating units while considering various operational constraints and market conditions.

To inform the development of the surrogate model, these simulations will involve varying different inputs and analyzing the resulting outputs. This comprehensive analysis will identify key patterns and relationships within the data, which will then be used to construct a surrogate model that accurately approximates Dispa-SET’s outcomes. The surrogate model will focus on replicating critical results related to load shedding and curtailment. Both aspects are crucial for effective power system simulation and management, as they significantly affect grid stability and the integration of renewable energy sources.

Through this analysis, it was determined that the key features influencing curtailment include the power capacity of wind and the transfer capacities. For load shedding, the most significant factors are the capacity ratio and the transfer capacities.