A Distributionally robust optimization model based on data mining for energy management of distribution network with renewable energy

Abstract

The uncertainty of renewable energy restricts their penetration in distributed network. Demand response (DR) provides a new scheme to improve the utilization rate of renewable energy. However, traditional stochastic optimization and robust optimization methods have some limitations dealing with uncertainty of renewable generation and demand response. This paper proposes a distributionally robust optimization model for distributed network energy management to improve the utilization rate of renewable energy. Firstly, a large number of historical data of user load and renewable generation are analyzed by K-means clustering method to obtain typical scenarios and their corresponding probability distribution. Then, the confidence set of probability distribution constrained by 1-norm and $\infty$ -norm is established to construct the distributionally robust optimization model for distributed network energy management. This model considers uncertain renewable energy and user load under the worst probability distribution, and dispatches user load and distributed generation in the condition of meeting security constraints and user power consumption constraints to search the optimal solution. Finally, columns and constraints generation (CCG) algorithm is proposed to solve the model, and the effectiveness of proposed model is verified based on simulation.