Transmission line maintenance is an important way to ensure highly reliable power supply.The adoption of intelligent power maintenance equipment to replace manual labor has become a trend.Transmission line maintenance equipment operates in the environment of high-altitude strong wind disturbance and strong electric interference for a long time，and its key components are easily damaged，which leads to failures.Therefore，it is significant to accurately diagnose the faults of intelligent power maintenance equipment for ensuring the safe and stable operation of the power grid.In this paper，spatial-temporal graph is created from the measured signals by short-time Fourier transform，and the fault diagnosis features of intelligent power maintenance equipment are extracted.The deep reinforcement learning framework is used to enhance the diagnosis process，and ChebyGCN is used to classify the fault features in different fault states and working conditions.This is the first time that DRL and ChebyGCN are jointly used for fault diagnosis of intelligent power maintenance equipment.Comprehensive experiments are carried out on a self-built dataset and a public dataset to verify the effectiveness of the proposed method，in which its diagnostic accuracy is 99.51% and 100%，respectively.Compared with common prediction models，the proposed method can achieve higher prediction accuracy.