Optimized Settings of Droop Parameters Using Stochastic Load Modeling for Effective DC Microgrids Operation

摘要

Droop control is a widely used technique for load sharing in dc microgrids (MGs). However, it has an inherent limitation that leads to voltage deviations at the dc bus. More importantly, its current-sharing performance is degraded due to voltage drops across transmission line impedances. Depending on the amount of these voltage drops over the dc bus, droop control leads to different current-sharing errors and dc bus voltage degradations. Therefore, MG topology and loading conditions (LCs) have a considerable effect on current-sharing accuracy and voltage regulation. In this paper, an optimization procedure accounting all necessary information is introduced to find optimal droop parameters for the droop mechanism. First, a proper optimization problem is defined with required constraints and a cost function formulated as a summation of the current-sharing errors and the voltage degradations occurring at various LCs. Then, an optimization tool named as particle swarm is implemented to converge a satisfactory solution. During the computations, the cost impact of the current-sharing error at each LC is weighted based on the probability of occurrence of that LC, which is obtained from the stochastic load model for the considered MG. The effectiveness of the optimal droop parameters is verified through a simulation performed on MATLAB/Simulink and a dc MG test bench is prototyped for experimental validation.