Abstract: This paper proposes a method for fault diagnosis and prediction of transmission and distribution lines based on big data analysis, utilizing the CNN-LSTM hybrid model and RMT-CNN technology to achieve accurate fault diagnosis and localization. Experiments demonstrate that this method can effectively enhance the operational reliability of power systems, reduce the risk of faults, and provide technical support for the stable operation of power systems.