•  
  •  
 

Keywords

distributed generation; low-voltage distribution station area; graph convolutional network; cloud-edge collaboration; phased voltage control

Abstract

In view of the voltage overrun problem triggered by the unsynchronized peak output period of distributed generation (DG) and heavy load period in low-voltage distribution station areas, a phased voltage control strategy for DG in station areas based on cloud-edge collaboration is proposed. First, the historical operation data of the station area collected by the intelligent fusion terminal is received through the cloud master station, and a short-term prediction model of voltage based on spectral graph theory and graph convolution network is constructed at the cloud layer; then, the intelligent fusion terminal at the edge layer makes short-term prediction of voltage by using the voltage prediction model sent down from the cloud layer and the real-time data of the station area. The impact of the distribution of active energy on the voltage is analyzed by taking into account the relatively large sense of resistance of the low-voltage distribution network, and the grid-connected inverter of DG performs control in three stages: reactive power coordination, active power reduction, and power recovery, based on the voltage prediction values sent down from the edge layer, quickly bringing the voltage back to the normal range; when the voltage prediction value has an error with the real-time voltage, the edge layer uploads the new dataset that has an error to the cloud master station to optimize the model, realizing the collaboration of the cloud and the edge layers under the real-time voltage control strategy. Finally, the proposed control strategy is analyzed through an example of a rural low-voltage distribution station area to deal with the problem of voltage overrun in the station area and verify the effectiveness of the proposed strategy.

DOI

10.19781/j.issn.1673-9140.2025.03.006

First Page

52

Last Page

60,122

Download

Share

COinS