Multi-Agent System Based Distributed Pattern Search Algorithm for Non-Convex Economic Load Dispatch in Smart Grid

Fangyuan Li,Jiahu Qin,Yu Kang

DOI: https://doi.org/10.1109/tpwrs.2018.2889989

IF: 7.326

2019-01-01

IEEE Transactions on Power Systems

Abstract:Since generation units usually subject to valve point effects and there is a trend of ever-growing number of distributed energy sources, distributed optimization algorithms for non-convex economic load dispatch are more desirable than the centralized ones. In this paper, a distributed pattern search algorithm (DPSA) for non-convex economic dispatch is developed. To develop such an algorithm, a flooding-based topology discovery algorithm (FBTDA) is proposed first. Theoretic analysis shows that the FBTDA algorithm converges in the least number of iterations compared with existing methods. Then, the DPSA algorithm is developed by integrating the original pattern search algorithm and the FBTDA algorithm under the framework of multi-agent systems. The DPSA algorithm has a rather weak requirement on the communication topology. Rigorous convergence analysis is provided as opposed to previous literatures where convergence property is often not investigated. The correctness and efficiency of the proposed algorithms are verified by numerical simulations, which also show that the DPSA algorithm can improve performance by actively choosing communication topology and integrating other global search methods. Comparisons with traditional methods show that besides the potential benefits arising from the absence of central nodes, the DPSA algorithm has comparable performance to these methods.