Series AC arc fault detection method based on spectrogram and deep residual network

The extended and excessive use of power equipment can hasten the aging of circuit cables, resulting in arc faults. The generation of arc fault will not only affect the performance of power equipment, but also bring about safety hazards. Therefore, it is necessary to detect arcing in circuits. This paper presents a framework for detecting series arc faults based on spectrogram and deep residual network. The problem of current signal detection can be converted into the problem of image recognition by this framework. In this framework, the current signal is converted into a spectrogram, which enables the characterisation of the current signal from a multi-domain perspective. Then, a deep residual network model is used to recognize the spectrogram and determine the type of arc fault. Finally, the current data is used to demonstrate the effectiveness and accuracy of the proposed method. The results show that the proposed method is able to achieve accurate arc fault detection with an accuracy of 97.50 %.