Abstract: In order to prevent the foreign matters in the process of coal mining and transportation from damaging the transportation equipment and production equipment, it is proposed to research a belt conveyor fault auxiliary recognition system based on machine vision deep learning in combination with the traditional belt conveyor detection system. Image preprocessing is carried out through the image algorithm library to enhance the detectability of the system for relevant information; the recognition network model obtained by in-depth learning training uses monitoring video to identify foreign objects, improve the accuracy of the system to identify foreign objects, and effectively improve the transport efficiency of the transport link. The test results indicate that the fault auxiliary identification system can ensure the normal operation of the transportation system in the fully mechanized mining face.