Fully Decentralized Robust Modelling and Optimization of Radial Distribution Networks Considering Uncertainties

Fully decentralized optimization of multi-agent distribution networks considering uncertainties is essential to improving user-side energy utilization efficiency and flexibility, whereas a contradictory centralized coordinator aware of system-level information is inevitably introduced in existing researches. To address the problem, this paper introduces the flexibility boundaries of nodes to express their adjustability under uncertainties and constructs a flexibility transition model to express their neighbourhood relationship. Besides, a robust interval power flow model is established to consider the stochastic impact of generation on nodal voltage through neighbourhood information exchange. Based on the above two models, the robust optimization problem is established to minimize the baseline operation cost and maximize the allowable generation limits of non-dispatchable renewable energy sources. The model involves no global information and is solved with alternating direction method of multipliers(ADMM) in a fully decentralized way. Case study on a modified IEEE 33-bus and a 118-bus system is presented and the proposed method is compared with conventional multi-level robust formulations. The results suggest the effectiveness and correctness of the newly proposed method.