Power Grid Fault Diagnosis Based on SSAE and CNN

The fault of power grid will cause serious personal safety problems and economic losses. It is very important to diagnose the power grid fault accurately and quickly. In order to improve the fault diagnosis accuracy for hybrid AC-DC power grid, this paper proposes a stacked sparse autoencoder-convolutional neural network method. The paper uses stacked sparse autoencoder (SSAE) to reduce the dimensionality of high dimensional data sets, and then uses convolutional neural network (CNN) to extracts data features to diagnose different line faults and different types of faults in the power grid. Finally, the effectiveness of the proposed method is validated by MATLAB simulation, and shows that the proposed method has a high accuracy to distinguish different faults.