Aviation Plug On-site Measurement and Fault Detection Method Based on Model Matching

This paper proposes a novel on-site measurement and fault detection method for aviation plugs based on model matching. Model matching means matching model aspect-graphs and image aspect-graphs. This method involves the use of image processing technology for fault diagnosis and solves the problem of fault diagnosis of a large number of aviation plugs temporarily docked at important equipment sites. The method adopts a hierarchical k-means clustering scheme and ends the clustering of each layer by using customized discriminating conditions. According to the identified plug attitude parameters, we just transform the plug matching template but not the plug image, and then calculate the difference between the transformed template pin coordinates and the plug pin coordinates. When the difference is greater than the threshold, it is determined that the pin has a fault. This method is more stable in detection. Proposed method does not transform the original image, so that the pixel error caused by the original image transforming can be avoided, and the experiment results on real aviation plug also proved this advantage.