A machine vision-based automatic inspection system for power station coal bunkers maintenance

Automation technologies can play a vital role in thermal power plants equipment maintenance, where coal bunker liner are critical parts in terms of functioning and safety. In this context, common issues involve the premature wear of fastening hardware (e.g. bolts) and lining edge warping. Current inspection approaches rely on manually driven visual detection of individual liners, resulting in low efficiency, low accuracy and prone to occupational health hazards. To overcome such drawbacks and to boost automated maintenance, this paper proposes a machine vision-based automatic inspection system endowed with a Cable-Driven Parallel Robot and a machine vision unit for the defect detection in coal bunkers. Considering the reflective characteristics of stainless-steel lining plate and the problems of uneven brightness and unequal focal length of images caused by camera motion during image acquisition, the proposed system utilizes a non-equidistant defect detection method based on improved template matching algorithm under diffuse reflection light source. A tailored light source unit is designed to preliminary reduce the reflection of the stainless-steel lining plate, allowing for high-quality images acquisition of the areas to be inspected. An automatic defect detection algorithm is developed to identify and locate the defects using the real-time spatial geometric position of the camera. The experimental results show that this method can effectively and efficiently detect a number of defects types in coal bunker lining plate. In this respect, compared with current manual inspection methods, the proposed approach can drastically reduce the inspection time whilst keeping an excellent detection accuracy capability.