Research on the Emergency Load Shedding Strategy for Transmission and Distribution Networks with High Penetration of Distributed Photovoltaic
With integration of high proportions of distributed photovoltaic and different levels of important loads into new power systems,the traditional emergency load shedding strategies have poor adaptability to increasingly tight interconnected power grids,leading to the disconnection of distributed photovoltaic or important loads,and subsequently causing more serious chain failures.This paper presents an emergency load shedding strategy for transmission and distribution networks with high-scale distributed photovoltaic.First,the extreme operating parameters of transmission and distribution networks are extracted and emergency scenarios set is constructed through risk assessment.Second,an event-based and response-based dual-drive load shedding strategy is proposed in the transmission network layer.The load shedding optimization model of the transmission network is established to calculate the load shedding amount of the distribution network in emergency scenarios,which is stored in the offline database for the event-driven module,and the random forest model based on the offline database is used for the response-driven module.Furthermore,at the distribution network layer,by analyzing the importance of each feeder,the optimal load shedding model is constructed with minimizing the severity of load shedding consequences as the goal.The interruptible load is taken as the first round of load shedding to find the optimal load shedding quantity and location.Finally,the simulation analysis of a 105-node system with high-scale distributed PV is carried out,comparing with the other two load shedding strategies,the results show that the proposed strategy can effectively reduce the risk of power network and reduce the impact of both distributed PV and important load offline on the system.
risk assessmentrandom foresthierarchical structureload sheddinginterruptible loadseverity of load shedding consequences
NETL
NSTL
万方数据
维普
