This paper proposes a hybrid efficient stochastic framework for optimal operation of automated distribution grids considering the social, economic and technical priorities. To this end, optimal feeder reconfiguration is regarded as a key solution for altering the network topology and improving the above targets. Considering the wide popularity and high penetration of the plug-in hybrid electric vehicles in the future grids, three different charging schemes including controlled charging, uncontrolled charging and smart charging are introduced and compared with each other. In addition, different types of non-dispatchable renewable power sources including wind turbines and solar panels are considered to let the analysis be more practical. In order to capture the uncertainties existing in the nature of the problem, a hybrid data driven approach based on machine learning and unscented transform is proposed. The machine learning based approach, i.e., support vector machine, will help to estimate the standard deviation of the uncertain parameters. In order to make a global search in the multi-objective space, a new optimization method based on flower pollination algorithm and a new three-stage modification method are introduced to solve the problem. The quality and capability of the proposed framework are examined on an IEEE test system.

 

Energy Vol 220

Navid Parsa, Bahman Bahmani-Firouzi, Taher Niknam

 

Link de acesso:

https://www.sciencedirect.com/science/article/pii/S0360544220328103