Partial discharge(PD) is the main insulation defect of gas insulated switchgear (GIS),when it occurs in GIS, the insulation will be deteriorated and breakdown will finally happen, so it is very important to detect and localize internal PD as early as possible, and then carry out defect inspection and repair work. This paper proposes a PD recognition and localization method based on data driven, which means threshold and correlation statistics analysis of time series by multiple sensors. In this paper, sensor’s location is coded firstly, then the spatial and electric position of sensors can be identified by machine, and cumulative discharge amplitude of chracteristic is constructed to distinguish stable PD. According to the correlation analysis of monitoring values of sensors, the propagation way of PD is judged, and whether or not it being an internal PD can be identified. For an internal PD, its position can be located by comparing the cumulative discharge amplitude. Verification by online monitoring data of Qianfan station also proves that the discharge phenomenon can be recognized effectively and the method is feasible.