Abstract: Partial discharge detection is one of the important methods for assessing the insulation status of vehicle-mounted high-voltage cables. Narrowband interference, as one of the most common interferences, causes the partial discharge signal to be submerged in noise, which makes it difficult to obtain an accurate partial discharge signal and brings difficulties to the assessment of the cable insulation state. A method based on singular spectrum analysis is introduced for suppressing the narrowband interference in the localized discharge signal of vehicle-mounted high-voltage cables. The method realizes the analysis and prediction of one-dimensional signals by singular spectrum analysis(SSA), which decomposes the trend term, period term and noise from the time series. In order to verify the effectiveness of the proposed method, an ideal localized discharge signal with narrow-band interference is simulated, and the proposed method is applied to noise reduction of the signal. At the same time, the validation experiments of real localized discharge signals are carried out in the laboratory. The noise reduction results show that the proposed method can effectively eliminate the narrow-band interference in the partical discharge signal of the vehicle-mounted high-voltage cable, which is of some significance for judging the insulation status of the vehicle-mounted cable. This study provides a new method for localized discharge signal extraction for on-board high-voltage cables, which has potential engineering applications for the assessment of insulation status of on-board cables.