A model-less control algorithm of DC microgrids based on feedback optimization

Abstract

This work addresses the problem of the optimal real-time control of a DC microgrid without relying on its corresponding network model. The main goal of such a controller is to keep the nodal network voltages within the regulatory limits while offering current sharing capability between the different controllable generators powering the DC microgrid. The proposed model-less methodology is based on feedback optimization, which takes advantage of the available real-time measurements to update the setpoints of the DC generation assets. The optimal control variables are determined in an iterative manner by applying a primal–dual saddle-point method, which guarantees appropriate convergence features. The paper details both centralized and distributed implementations which are compared through simulations. The results evidence a good dynamic performance and an optimal steady-state operation as the proposed control algorithm converges to the solution provided by a conventional model-based Optimal Power Flow.