Strengthening Resilience of Electric Power Distribution Systems against Natural Disasters: A Perspective from Machine Learning and Optimization

Abstract

The resilience issues in the power system have attracted increasing attention worldwide, especially for the distribution systems that suffer from extreme weather events, such as hurricanes and wildfire. In this dissertation, several novel algorithms, such as safe reinforcement learning algorithm and risk-constrained adaptive robust optimization approach are proposed to provide resilient proactive scheduling strategies, emergency response strategy and restoration strategy for central controllers in the distribution system. Microgrids are proposed to serve as single entities from the perspective of the distribution system operator to enhance the resilience of the distribution system, reduce the distribution system operator’s control burden and improve the power quality of the distribution system. Uncertainties related with the extreme weather events such as power generation of distributed generators, intermittent load demand, point-of-common-coupling/tie-line conditions, and trend/trace of the extreme weather event are tackled through a combination of optimization approaches, artificial intelligence algorithms and risk management methods. Extensive simulation results based on real-world data sets show that the proposed novel algorithms based proactive scheduling strategies, emergence response strategy and restoration strategy can ensure the resilience of the distribution system in a real-world environment.