As an indispensable part of realizing mechanization and automation of handling system in ports, metallurgy, chemical, thermal power, building materials and other fields, conveying equipment faults account for 50% of delivery system faults. In order to improve the maintenance mode of conveying equipment based on manual inspection and record, this paper deals with the research and application of computer vision and predictive maintenance in health management of conveying equipment. Firstly, based on the research of mechanical structure of conveying equipment drive unit, failure cause analysis of conveying equipment and overview of common detection techniques were introduced. Secondly, an health management system based on computer vision and predictive maintenance was designed, the computer vision model was built based on Pangu large-sized model and trained by artificial intelligence training platform, and StressWave analysis method was applied for predictive maintenance of conveying equipment driving unit. In the application case of main belt of blast furnace and the key belt of sintering in the iron making process, this health management system issue alarms accurately and timely, minimize the impact of unplanned downtime on the production process, resulting in significant economic and social benefits.