Abstract— This paper proposes an adaptive and distributed secondary voltage control for microgrids with inverter-based distributed generators (DG). The proposed control is fully adaptive and does not require the information of DG parameters. Neural networks are used to compensate for the uncertainties caused by the unknown dynamics of DGs. The controller structure is fully distributed such that each DG only requires its own information and the information of its neighbors on the communication network. Therefore, this secondary control is associated with a sparse communication network. The effectiveness of the proposed methodology is verified for different loading, outage, and islanding scenarios as well as variable communication structures in a microgrid setup. Index Terms—Adaptive control, distributed cooperative control, distributed generation, inverters, microgrid, multi-agent systems. I. INTRODUCTION icrogridsare small-scale power systems that facilitate the effective integration of distributed generators (DG) [1]- [5]. They can disconnect from the main grid and enter islanded operation due to the preplanned scheduling or unplanned disturbances. Once islanded, the primary control is applied to maintain the DGs’ voltage stability. However, even in the presence of the primary control, DG voltages can still deviate from their nominal values. Therefore, an additional control level, namely the secondary control, is required to restore them [6]-[10]. Up to this point, two main secondary control structures have been proposed in the literature. The conventional secondary control of microgrids assumes a centralized structure that requires a complex communication network [11]-[15], in some cases, with two-way communication links. This exposes a single point-of-failure that could reduce the system reliability. Alternatively, distributed control structures [16]-[18], with a This work was supported in part by the NSF under Grant Numbers ECCS- 1137354, ECCS-1128050, and Office of Naval Research under award N000141410718. A. Bidram, A. Davoudi, and F. L. Lewis are with the University of Texas at Arlington Research Institute, University of Texas at Arlington, 7300 Jack Newell Blvd. S., Ft. Worth, TX 76118 (e-mail:ali.bidram@mavs.uta.edu; davoudi@uta.edu; lewis@uta.edu). S. S. Ge is with the School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 610054, China, and also with the Social Robotics Laboratory, Interactive Digital Media Institute, and Department of Electrical and Computer Engineering, National University of Singapore 119260 (e-mail: samge@nus.edu.sg). sparse communication network, can be used for the secondary control [19]. The distributed secondary control obviates the requirement for a central controller, and is more reliable. Existing efforts (e.g., [19]), however, have been geared toward microgrid systems with fixed, known system parameters. In practice, it is desirable to have an adaptive [20]-[24] control paradigm that compensates for the nonlinear and uncertain dynamics of DGs. The controller should be fully independent of the DG parameters, and its performance should not deteriorate by the change in DG parameters (e.g., due to aging and thermal effects). As opposed to [19], this paper proposes an adaptive and distributed secondary voltage control that satisfies the above conditions in a distributed fashion. Linear-in-parameter neural networks (NN) are used to compensate for the uncertainties caused by the unknown dynamics of DGs [25]-[29]. The microgrid is considered as a multi-agent system with DGs as its agents. The secondary voltage control is formulated as a tracking synchronization problem of the resulting multi-agent systems. DGs can communicate with each other through a communication network modeled by a directed graph (digraph). The Lyapunov technique is adopted to derive fully distributed control protocols for each DG. These control protocols are formed based on the NN adaptive weights, which are calculated in realtime. The salient features of the proposed methodology are: x A distributed control method is proposed to solve the tracking synchronization problem for multi-agent systems with unknown nonlinear dynamics. It is used to design an adaptive and distributed secondary voltage control. x This secondary voltage control is adaptive and does not require the information of DG parameters. x Each DG requires its own information and the information of its neighbors on the communication digraph; i.e., the proposed method is fully distributed. Therefore, a sparse communication structure can be utilized. The paper is organized as follows: Section II discusses the primary and secondary control levels in a microgrid control hierarchy. In Section III, the preliminary of graph theory is presented. In Section IV, neural networks are used to design an adaptive secondary voltage control based on the distributed cooperative control. The proposed control is verified in Section V using a microgrid test system. Section VI concludes the paper. Distributed Adaptive Voltage Control of Inverter-based Microgrids M Ali Bidram, Graduate Student Member, IEEE, Ali Davoudi, Member, IEEE, Frank L. Lewis, Fellow, IEEE, and Shuzhi Sam Ge, Fellow, IEEE Digital Object Identifier 10.1109/TEC.2014.2359934 0885-8969 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications standards/publications/rights/index.html for more information. http://ieeexplore.ieee.org/Xplore