An identification method for weak links in distribution network feeders based on RF-LSTM under extreme weather

Abstract

An identification method for weak links in distribution network feeders based on RF （random forest） and LSTM （long and short-term memory network） is proposed to enhance the active defense of the distribution networks against extreme weather and improve power supply reliability. First， the short-time prediction of power flow under extreme weather is carried out based on historical operation data using LSTM， and the prediction results and weather forecast information are used as input parameters of the fault prediction model. Afterward， the feeder fault prediction model of the distribution networks under extreme weather is constructed using the RF algorithm， and the historical data are learned and trained. Finally， the short-term power flow data， meteorological parameters， and grid frame data obtained from the LSTM prediction are input into the RF prediction model and calculated. The fault probability of the distribution network feeders is predicted， and their weakness is classified to identify the weak links in the distribution grid feeders under extreme weather. The simulation results show that the method can accurately determine the weak links in distribution network feeders and can serve as a practical reference for improving distribution networks' active operation and maintenance capability.

Keywords

• distribution network
• extreme weather
• rf-lstm algorithm
• fault prediction
• weak links identification