Efficient identification of critical cases that lead to instability in future power systems

This PhD project will enable efficient identification of critical operating conditions and disturbances that might lead to instability on highly complex and uncertain power systems. The initiating disturbance will be also treated as an uncertainty which will address HILP events since biased sampling will be performed towards events that might lead to instability irrelevant of their probability. A combination of game theory with deep neural networks to guide the simulations is expected to provide significant improvements in the efficiency of current methods and tackle this very significant problem. Genetic algorithms have also shown promising results; outperforming machine learning in some cases when used in the game domain to explore large search spaces, a very similar problem faced in the context of this PhD.
Identification of critical operating conditions and disturbances that might lead to instability on highly complex and uncertain power systems. A combination of game theory with deep neural networks to guide the simulations is expected to provide significant improvements in the efficiency of current methods.