Distributed Optimization Method For Intelligent Control Of DC Microgrids

摘要

Most of the distributed energy resources, loads and energy storage systems in a DC microgrid are equipped with power electronic converters. With the integration of advanced power electronics devices, a microgrid is able to utilize a broader range of technologies in its design and operation. A key feature of power electronic converter based systems is the ability to direct energy flow within a system with their coordinated operation. A system level control is needed for coordination where converters execute reference points dictated by a system-level control in order to achieve system level goals. System goals can be expressed as a cost function solved by a real-time optimization algorithm. This work develops a framework for the coordinated operation of converters with a distributed optimization method for use in a real-time system-level control system.;In order to validate the optimization based control method developed in this research, a simplified shipboard DC power distribution system is used for case studies. It is an isolated microgrid with converters between all sources of energy and the main buses as well as between all load centers and the main buses. The example cost function used in the study minimizes distribution losses in the DC power system. Initially, the optimization problem is solved using a centralized method in order to provide a baseline for evaluating other schemes. Primal-dual interior point method is applied successfully to provide optimal operating points. The centralized structure relies on one central controller to support the entire system control such that the system is vulnerable to single points of failure and not easily expandable.;To address the robustness and expandability shortcomings, a distributed coordinating optimization algorithm is developed. The coupling constraints formed by nodal current balance result in control variable coupling, therefore, techniques are required to perform an appropriate decomposition. The main task of this dissertation is to develop a practical distributed algorithm via the decomposition of the optimization problem. The method developed here combines dual decomposition and Alternating Direction Method of Multipliers (ADMM) together. This is an iterative based method. By utilizing a decomposition method, the microgrid is partitioned into multiple subsystems. The global target is achieved by interaction of the subsystems which operate on local information. The solutions from the decomposition method and centralized method are compared in diagrams and in numbers using the shipboard DC microgrid test system. Results show that the numerical results from both methods match closely. Analysis of the effect of the number of microgrid subsystem partitions on convergence speed of the decomposition method is also performed.