Topology-Based Stabilization of Islanded Microgrids With Multiple Tie Switches Under Operational Variations: A Neural Control Approach

Topological flexibility of islanded microgrids (IMG) has recently shown significant potential for system stabilization. This paper proposes a neural approach for topology control of IMGs, with the objective of stabilizing the IMG with an arbitrary number of controllable lines and variable system operating conditions. The stability and stabilizability of IMGs are both assessed to determine the activation status of the topology controller which stabilizes the IMG through state-dependent switching of meticulously selected controllable lines. The underlying problem of the topology control is to find the real-time computable state-dependent line switching control law and stabilizability margin (SM) formulation. To address this problem, we establish the associated theoretical foundation within the framework of multiple Lyapunov-like functions of switched systems; then, we develop the neural line switching control law and neural SM formulation, both utilizing off-line trained neural networks for real-time computation irrespective of variations in the system's operation conditions. Lastly, simulation results on a 14-bus IMG system and an 85-bus IMG system demonstrate the efficacy of the proposed topology control approach.