Abstract: Aiming at the problems of low accuracy, easy to be affected by environment and high monitoring cost, a fault diagnosis method of conveyor idlers based on the infrared thermal imaging technology is proposed in this paper. The infrared thermal imager is mounted on the orbital inspection robot to carry out real-time monitoring of the temperature of belt conveyor idlers. In MATLAB, image processing operations such as the Wiener filtering, grayscale processing, median filtering, edge extraction, etc. are carried out for the collected infrared images. The human-computer interaction interface is designed in LabVIEW, which can display a series of information such as original infrared image, edge image, the maximum temperature and fault location of the faulty idler in real time. The co-simulation results show that the method can find the fault roller in time and report the location accurately, which improves the intellectualization degree of coal mine.