Improved ADP-Based Limited-Iterative Decentralized Algorithm for Optimal Power Flow of Networked Microgrids

In this paper, an improved approximate dynamic programming (IADP)-based limited-iterative decentralized algorithm is proposed for solving the optimal power flow (OPF) problem of networked microgrids (NMGs). Unlike traditional approximate dynamic programming (ADP) algorithms, which need to iteratively update the value function at a large number of states and scenarios, IADP can directly descript the value function by an infinite-dimensional Bellman's equation. Considering that this equation is intractable in theory, an infinite-dimensional projection technique is further proposed to transform it into a finite-dimensional problem. The IADP algorithm is applied to solve the decentralized OPF problem of NMGs. Different from most existing decentralized algorithms, the proposed algorithm can directly obtain the solution while omitting iterations between networks. This helps to enhance the computation efficiency and reduce the communication burden during the decentralized optimization process. Case studies on the 49-bus NMGs and the 265-bus NMGs illustrate the effectiveness of the proposed algorithm.