Fault Prediction of Street Lamp Power Distribution System Based on Sliding Time Window with Attenuation of AR-ELM

Abstract

Aiming at the street lamp power distribution system, a fault prediction method based on the improved online limit learning machine is designed, which is integrated into the AR Model to enhance the temporal correlation analysis of the ELM model, into the attenuation sliding time window to speed up the efficiency and accuracy of the model learning. The application showed that the method can prevent a wide range of lines and distribution problems caused by the street lamp power distribution system failure.