Distributed Model Predictive Voltage Control for Distribution Grid Based on Relaxation and Successive Distributed Decomposition

Abstract

Advanced control techniques for modern distribution grids are becoming fundamental for the reduction of grid reinforcements while maintaining network performances. In particular, as shown in literature, distributed algorithms for voltage control have gained much interests in comparison with the typical centralized formulation for its feasibility. Distributed model predictive control (MPC) is shown to optimally manage the Distributed Generators (DGs) over time and it can be implemented locally at the controllable sources. For this purpose, this paper adopts a distributed algorithm recently presented in the literature for solving a constraint-coupled optimization problem for model predictive voltage control. A detailed reformulation of the original MPC problem for the specific application is presented. Besides, this paper provides a calculation of the convergence limit for the value of the iteration step size of the algorithm, supported by numerical results. The proposed distributed solution of model predictive voltage control is compared with a centralized formulation via numerical simulation in terms of the percentage of error with the centralized solution and number of iteration for the convergence.

Keywords

• Voltage control
• distributed optimization
• smart grids
• networked control systems