The pulse current method and the secondary pulse method can locate the power cable faults effectively, but they rely on manual judgement of the breakdown discharged waveform, which is inefficient. To identify and locate faults automatically, this paper proposes an intelligent cable fault location method. Firstly, the models of cable fault location are studied and the characteristics of fault waveforms are analyzed. Then a Grid Search-Support Vector Machine (GS-SVM) based algorithm for the recognition of breakdown discharged waveform is proposed. And the fault distance of cable can be calculated directly after the recognition. Finally, the field test data are used to verify that the proposed method can not only ensure the accuracy, but also improve the efficiency of locating cable faults.