A remote barrier localization method for intelligent substations using deep belief networks

In the remote operation of intelligent substations， failures often occur due to device or network malfunctions， or inconsistencies in information， consequently hampering operation and maintenance efficiency. To address this， the paper proposes a remote barrier localization method for intelligent substations using deep belief networks （DBNs）. Firstly， the whole process of remote control in an intelligent substation is decomposed to categorize the failures into three main types. Then， DBN-Softmax classifiers are introduced to extract features and train the classification model respectively， building a remote barrier localization model. Finally， failure types and telemetry information are used to precisely locate the faults. Engineering examples have shown effective validation of the method， including its merits of short computation time and high localization accuracy.