Abstract
This paper investigates the state estimation problem for an islanded DC microgrid. The microgrid consists of energy storage units, as well as wind and solar generation units, all designed to supply power to a variable DC load, specifically an electric vehicle charging station. To ensure accurate state estimation, a third-order model is utilized for the battery storage system. A DC/DC boost converter is implemented for the photovoltaic system, resulting in a second-order state-space model. The wind generation unit, which incorporates a three-phase synchronous generator, is represented by a fourth-order nonlinear state-space model. Consequently, the entire microgrid is described by a ninth-order nonlinear state-space model, which is simplified by excluding two states from the battery model. This study also includes a comprehensive analysis of the load's dynamic modeling and behavior. A second-order state-space model is derived for the electric vehicle charging station as the load. A variant of the Kalman filter is employed as the state estimation algorithm, which is model-based. The algorithm is implemented to estimate the unknown or unmeasurable states of the microgrid, incorporating known inputs including load, solar irradiation, wind speed, and photovoltaic panel temperature. The algorithm employs an unscented transform to eliminate the need for system linearization, with the Kalman filter playing a pivotal role in the estimation process. The robustness of this convergence is evaluated under three scenarios: first, when encountering sudden load variations; second, when improperly initializing the estimation process; and third, during abrupt changes and noisy measurements in solar irradiation, one of the system inputs. Simulations conducted in MATLAB/SIMULINK demonstrate that the algorithm's estimations converge to actual values with a satisfactory level of accuracy and precision within a desirable time interval.