Abstract: In this paper, a method of multi-source partial discharge localization based on the frequency domain threshold processing generalized cross-correlation (FDTP-GCC) and the spatial screening for the multi-source partial discharge localization problem of the power equipment is proposed. In order to realize the same-source time difference extraction of multi-source partial discharge ultrasonic signals, the FDTP-GCC algorithm is suggested in this paper. This algorithm is directly applied to the original signals collected by ultrasonic sensing devices, excluding the interfering frequency bands in the frequency domain by setting the amplitude spectrum threshold. Then the amplitude spectrum is normalized to highlight the impulse function characteristics of the multi-source signal time difference, directly obtaining the time difference of the corresponding same-source signals. The simulation tests verify the high accuracy, stability and noise immunity of the FDTP-GCC algorithm for the time difference estimation. In order to solve the problem of the same-source time difference separation brought by the FDTP-GCC algorithm, this paper puts forward a multi-source partial discharge localization based on the spatial screening and analyzes the localization performance of the spatial screening using the analytical solution-based algorithm and the intelligent algorithm. Finally, a localization test is conducted using the dual partial discharge sources, and the experimental results show that the time difference errors extracted by the method in this paper are all within 1μs, and the localization errors of the two partial discharge sources are 20.53 cm and 4.93 cm.