This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS 1 Decentralized Economic Dispatch in Smart Grids by Self-Organizing Dynamic Agents V. Loia, Senior Member, IEEE, and A. Vaccaro, Senior Member, IEEE Abstract —In this paper, we propose a decentralized and self- organizing solution framework aimed at addressing economic dispatch (ED) analysis in a distributed scenario. In particular we will demonstrate that, under some hypotheses, the solution of the ED analysis can be obtained by computing proper weighted averages of the variable of interests. To compute these global quantities we propose the deployment of a network of cooperative dynamic agents solving distributed average consensus problems. Thanks to this decentralized/nonhierarchical paradigm, all the basic operations needed to solve the economic dispatch problem could be easily processed by the agents. Simulation results obtained on the 118 and 300 bus IEEE test networks are presented and discussed in order to prove the effectiveness of the proposed framework. Index Terms—Intelligent systems, optimization methods, power generation dispatch, smart grids. I. Introduction E CONOMIC dispatch (ED) analysis is one of the most fundamental and heavily used tools in power systems studies. It aims at scheduling the committed generating unit outputs so as to meet the required load demand at minimum operating cost while satisfying all unit and system constraints [1]. From a mathematical point of view, ED analysis can be for- malized as a nonlinear constrained optimization problem [2]. To solve this problem, a large number of solution algorithms have been proposed in the literature. They include iterative numerical procedures based on the Lagrange multipliers theory [3], dynamic programming [1], evolutionary algorithms [4], and heuristic techniques [5]. Although these solution strategies offers considerable in- sight into the important role played by modern optimization techniques in ED analysis, their application asks for the deployment of a data fusion center acquiring and processing all the power system measurements. A debate on the adequacy of this hierarchical control paradigm in the context of the modern smart grids has been recently afforded in the power systems research community. Manuscript received June 20, 2012; revised March 18, 2013; accepted March 23, 2013. This paper was recommended by Associate Editor W. Pedrycz. V. Loia is with the Research Consortium on Intelligent Software Agent Technologies-CORISA, Department of Informatics, University of Salerno, Fisciano 84084, Italy (e-mail: loia@unisa.it). A. Vaccaro is with the Department of Engineering, University of Sannio, Benevento 82100, Italy (e-mail: vaccaro@unisannio.it). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TSMC.2013.2258909 In particular, it is expected that the large scale deployment of the smart grids paradigm will massively increase the data exchange rate leading centralized control architectures to becoming rapidly saturated. Consequently the streams of data acquired by distributed grid sensors could not provide system operators with the necessary information to act on in appropriate time frames. Even in the presence of advanced tools aimed at converting the data into information, the smart grid operator must face the following challenges [6], [7]: 1) communication bottlenecks; 2) complex control and optimization problems; 3) growing of energy management systems complexity; 4) centralized infrastructures can be a security target. To address these issues, smart grid researchers are reviewing the design criteria and assumptions concerning the scalability, reliability, heterogeneity and manageability of power systems control architectures [8], [9]. These research works conjec- tured that hierarchical control paradigms could be not af- fordable in addressing the increasing network complexity and the massive pervasion of distributed generators characterizing modern smart grids [10]–[12]. In this context, the research for distributed multi-agents optimization paradigms has been identified as the most promising enabling technology. This is mainly due to the successful application of decentralized and cooperative agents’ networks in enhancing operational effectiveness of complex systems [13]–[16]. Armed with such a vision, several papers outlined the important role played by multi-agent systems in addressing many smart grid control problems as far as optimal power flow studies [11], economic dispatch [10], load restoration [17] and voltage regulation [18] are concerned. These papers demon- strated that distributed control architectures could improve the power systems effectiveness by reducing dependencies and enhancing the system ability to remain in operation after disturbances and/or equipment loss [17]. Besides, if properly designed, they are characterized by stabilizing and self-healing proprieties. In our opinion, these features could be highly beneficial in addressing ED analysis in smart grids not only for parallelizing the solution algorithm at a global scale, but also for distributing responsibility among the distributed power generators. Be- sides, we strongly believe that in this domain the research for rigorous tools and scientific methodologies aimed at defining a fully decentralized and self-organizing control architecture is still embryonic and need to be researched. 2168-2216/$31.00 c  2013 IEEE