Abstract: For uncertainty of line fault location, current series fault arc detection methods are mainly based on current signal analysis. By comparing current waveforms before and after series arc fault under different loads, characteristics and regularities of series fault arc current were obtained. Taking current signal of series fault arc as research object, three kinds of series fault arc detection methods were introduced which use Hilbert-Huang transform, information entropy and short-time Fourier transform and wavelet approximate entropy and support vector machine respectively. Extraction processes of fault arc feature with different detection methods were summarized, and advantages and disadvantages of the three methods were compared. A view was pointed out that the detection method based on Hilbert-Huang transform and the one based on information entropy and short-time Fourier transform can effectively extract time-frequency characteristics of fault arc, and series fault arc can be identified according to proper threshold of time-frequency spectrum amplitude with low accuracy and real-time performance. The detection method based on wavelet approximate entropy and support vector machine can directly extract approximate entropy as input of support vector machine to detect series fault arc with higher accuracy and real-time performance, which is more suitable for coal mine.