Reinforcement learning approach for congestion management and cascading failure prevention with experimental application

(Deep) Reinforcement learning for electric power system control and related problems: A short review and perspectives

“…Q-learning was suggested in [57] to determine optimal control of active power generations for preventing cascading failure and blackout in smart grids. This approach belongs to subsystem level controls and considers single line outages (termed N-1 contingency in power system literature) and two consecutive line outages (termed N-1-1).…”

A Review of Neural Network Based Machine Learning Approaches for Rotor Angle Stability Control

“…It is shown in literature that, the WAM temporal information can further be used in WAC designs to perform real-time transient stability enhancement, which can improve the power transfer capability of a transmission system and prevent the system from generation or load disconnection, or catastrophic failure following a sequence of disturbances in the system. Article [96] have used the RL for preventing cascading failure (CF) and blackout in smart grids by acting on the output power of the generators in real-time. This article makes use of the state-action policy update feature of RL algorithm, as it can learn from interactions with the system.…”

Real-time security margin control using deep reinforcement learning

“…DRL-based implementations adapted for preventive control found in the literature are few and in [13], the authors argue that the reason may be because these control problems have traditionally been formulated as static optimization problems. One example of an RL-based implementation for preventive control is presented in [21], which aimed to determine the optimal control of active power generation for preventing cascading failures and blackouts. However, the method was based on a tabular form of Q-learning, which in general is not suited to handle large state and action spaces.…”