A robust dynamic state estimation for droop controlled islanded microgrids

This paper proposes a nonlinear particle filter (PF) method for dynamic state estimation in droop-controlled islanded microgrids (IMGs). The PFs are normally applied to systems that (1) have highly nonlinear system dynamics, and (2) do not require the additive process or observation noise to be Gaussian. This flexibility allows the PFs to handle noisy measurements from a range of varied distributions, thereby increasing its robustness. To that end, a nonlinear dynamic state model has been developed in this work for droop-controlled IMGs. Additive noise has been incorporated into the state model to account for the error in its accuracy. Monte Carlo simulations have been conducted to verify that the PF accurately tracks the IMG state variables in spite of using significantly corrupted state and observation values. A comparison between the PF and unscented Kalman filter (UKF) has been carried out to test the effectiveness and robustness of the proposed methodology.