Privacy-Preserving Collaborative Assessment of Multi-Day Operation Security in Distribution Systems with High Levels of PV Penetration

A distribution system with high levels of photovoltaic (PV) penetration is greatly affected by possible prolonged abnormal weather events that may last for multi-days and thus invite multi-day coupling insecure operation issues. Also concerning that the utility distribution grid (UDG) and microgrids (MGs) in a distribution system are owned and managed by different stockholders, the paper presents a privacy-preserving collaborative assessment method for the system's multi-day operational security. First, we characterize the multi-day evolution of PVs' and loads' forecast uncertainties by the customized martingale-like process. Then, we embed gradually increasing uncertainties in the coming days into the multi-day collaborative assessment model, whose optimistic and pessimistic decisions comprehensively indicate the respective normal, abnormal, or emergency states of networked UDG and MGs. Further, we formulate unified dual cuts and feasibility restoration cuts in the pessimistic case and optimistic case with respect to the mixed integer nature of UDG and MGs' assessment model, by which the collaborative decisions can be made in a decentralized way without disclosing privacy. The optimal convergence of decentralized decision-making and the effectiveness of relaxation for binary variables are proven in mathematics. Finally, the experiment demonstrates that on the promise of privacy preservation, the proposed method is not only correctness and timeliness but also scalable and robust.