Abstract: To address the high complexity of generating fault scenario data in power systems, this paper proposes a fault scenario generation strategy based on VMD-ConTimeGAN. Firstly, a VMD-based operation retrieval method is proposed to collect power system operation data. Secondly, ConTimeGAN is designed to capture the intrinsic patterns of time series data, and the training strategy of GAN is introduced. Finally, the stability of the data generated by the model is ensured, and the data set is integrated after evaluation.