Robust optimal capacity planning of grid-connected microgrid considering energy management under multi-dimensional uncertainties

Microgrid is considered an efficient paradigm for managing the massive number of distributed renewable gen-eration and storage facilities. The optimal microgrid capacity planning is a non-trivial task due to the impact of randomness and uncertainties of renewable generation sources, and the adopted energy management strategies. In this paper, an optimal capacity planning model for the grid-connected microgrid is developed fully considering the renewable generation uncertainties through efficient scenario generation and reduction based on the deep convolutional generative adversarial network (DCGAN) and improved k-medoids clustering algorithm, as well as the microgrid energy management strategy. The proposed solution optimizes the capacity planning for the maximization of renewable energy utilization efficiency, and minimizes the economic cost and carbon emissions. The proposed solution is assessed using a case study of a microgrid (MG) project in northern China through a comparative study and the numerical results confirm the cost-effectiveness of the proposed solution.