Linearized-Trajectory Model-Predictive Controller for Improving Microgrid Short-Term Stability With Real-Time Implementation

Abstract

A model-predictive controller (MPC) for synchronous generator-based microgrids is introduced to enhance voltage, frequency, and transient stability in fast timescales. The centralized controller uses mathematical models of power system dynamics to predict the evolution of system states over a finite horizon based on information from local state observers and relays. The MPC dynamically adjusts the setpoints of existing primary controls to maximize system stability. To address the computational challenges of a real-time MPC implementation due to nonlinear power system dynamics and short sampling intervals, a trajectory linearization technique is applied to the MPC formulation. The proposed controller is validated via a hardware-in-the-loop (HIL) testbed. The testbed includes an implementation of the controller in hardware, which interfaces with a high-fidelity microgrid simulation model running in a realtime simulator. State observers based on the Extended Kalman Filter (EKF), which provide the controller with estimates of the system state, are also implemented. The HIL testbed is used to verify the real-time operation of the controller and evaluate its performance under modeling uncertainties.