Abnormal State Identification of Capacitor Voltage Transformer Based on DA-Xception Algorithm
In order to accurately online identify whether a capacitor voltage transformer is undergoing abnormal states,a DA-Xception algorithm integrating DANet dual attention mechanism is proposed.This algorithm integrates the dual atten-tion of DANet into the output stream stage of the Xception algorithm,learning the interdependence between extracted fea-ture channel information and spatial information,and obtaining effective feature results that are more consistent with the contextual relationship of the target feature.The experimental results show that the DA-Xception algorithm can efficiently and accurately identify abnormal states of capacitor voltage transformers with an accuracy of 97.7%.
capacitor voltage transformerabnormal identificationXceptionDANetdual attention
NETL
NSTL
万方数据
