Consensus-Based Distributed Optimal Energy Management With Less Communication in a Microgrid

Abstract

In this paper, to reduce required capacities for information exchanges in microgrids, a novel distributed eventbased algorithm is proposed for optimal energy management in a microgrid. Aiming at optimally scheduling the energy supplier's generation, an objective function is formulated to minimize the total cost of maintaining the supply-demand balance with considering power losses. Regarding each participant as an agent, the proposed algorithm is implemented in a distributed manner based on a multi-agent system framework. Therefore, each agent only exchanges information with its neighbours through a local network. Additionally, comparing with the periodical communication of sampled-data mechanisms, the adopted event-based scheme achieves satisfactory performance by using significantly less communication between participants. As a result, it further facilitates the development of networked microgrids. Furthermore, concerning the privacy of participants, the proposed algorithm is implemented without exposing owners' private preferences. The effectiveness of the proposed distributed algorithm is validated through several simulation studies.

An exponentially convergent distributed algorithm for resource allocation problem

“… where denote the lower and upper resource bound of generator i , respectively. Furthermore, the following assumption is needed for the following designed distributed algorithm [36, 37, 42].…”

“…Moreover, on the basis of saddle point dynamics and consensus of multi‐agent systems, [38, 39] proposed a distributed algorithm for the resource allocation problem. [37, 42] proposed the distributed projected algorithm for solving the resource allocation problem. All the above algorithms achieve the optimal solution by establishing a differential equation.…”

“…Convergence speed comparison among algorithm (8), the penalty function method of first‐order in [35], the penalty function method of second‐order in [36], the saddle point dynamics and consensus algorithm in [38], the barrier function method in [32] and the projected algorithm in [37, 42]. [Color figure can be viewed at wileyonlinelibrary.com]…”

“…Furthermore, [32, 34–36, 38, 39] could only deal with the one‐dimension resource allocation problem while our proposed algorithm can be extended to multidimensional resource allocation problem. Second, in comparison with [37, 41–44], the proposed algorithm obtains the output of each agent through a static projection function instead of a differential equation, which is inspired by our previous paper [14]. Compared to the above dynamic projected method, in which the output is achieved by using the dynamic projected output feedback design, Our proposed algorithm is more sensitive to the Lagrange multiplier, which leads to faster convergence.…”

Robust optimal operation of future power systems under uncertainties

“…Power-sharing frameworks in MMG systems promote microgrid operators (MGOs) of the constituent MGs to optimize their local resources and cooperatively contribute to the MMG energy and reserve markets via the coordinating entity. Furthermore, this framework is also privacy-preserving as it only requires the individual MGOs to communicate their supply/demand bidding information [16][17][18][85][86][87][88][89][90][91][92][93][94][95].…”