Robust Optimal Scheduling of Distribution Networks Considering the Tail Correlation of Power and Load

Constructing reasonable uncertainty sets is a prerequisite for applying robust optimization in power system scheduling problems. To enhance the precision of the uncertainty set, this paper proposes constructing uncertainty sets based on tail correlation, jointly characterizes the power-load uncertainty in the system, and applies it to the optimization and scheduling of the distribution network. First, this paper proposed a specific definition of the generalized extreme value theory which is extended to the multivariate form. On this basis, the concept of tail correlation is introduced, and the tail correlation between uncertain variables on the source and load sides is calculated and analyzed. The extreme joint distribution of uncertain variables with strong correlation is fitted, and the percentile function of the joint distribution is taken to obtain the sample to construct the uncertainty set. An actual distribution network in Lu'an City, Anhui Province, is used as an example to verify the proposed approach. The results indicate that this method is able to precisely depict the uncertainty of the system, guarantee reliability, and decrease the conservatism of the scheduling plan.