A Novel Model Predictive Controller for Distributed Generation in Isolated Microgrids—Part II: Model Predictive Controller Implementation

Abstract

To realize operation automation in remote islanded microgrids, a model predictive control (MPC)-based distributed generation (DG) controller is proposed in Part 2 of this article. The developed data-driven predictive model in Part 1 of this article is implemented in Part 2 to realize the proposed MPC controller. The controller does not incorporate any tunable coefficients, and thus, is not sensitive to the system variations. Kalman filter-based state observer rejects measurement noises and updates the system model under varying system operation. The KWIK optimizer is used to solve the MPC's constrained quadratic programming problem as it ensures a guaranteed convergence. The proposed controller is smaller in size and does not require any intra-DG communication network. It also ensures proportional reactive power sharing despite feeder line impedance mismatch without requiring a secondary controller and a virtual impedance loop. The developed MPC controller is validated through case and sensitivity studies, showing excellent performance when compared with existing methods.