| Abstract:The occurrence of defects in the crimping of hard alloy tension wire clamps leads to increased contact resistance, local overheating, and even wire breakage accidents in transmission lines, posing a threat to the safety of power grid operation. To this end, a method for detecting the crimping defects of transmission line hard alloy tension clamps based on CenterNet algorithm is studied. Using X-DR (X-ray digital imaging) technology to obtain images of the tension clamp crimping of hard alloy transmission lines, and implementing homomorphic filtering processing. In the X-DR image of the tension clamp, the global contrast values of the hue, saturation, and brightness of each pixel are calculated to locate the significant areas of compression, thereby reducing the detection range of the CenterNet algorithm. Input the salient region images into the CenterNet algorithm, predict the probability of each type of crimping defect, and label the width and height to achieve the detection of crimping defects in transmission line hard alloy tension clamps. The results showed that the detection method based on CenterNet algorithm exhibited a high hit index (greater than 0.9) when facing four different types of crimping defects; In addition, the ROC curve of the method is closer to the upper left corner, indicating a lower risk of missed and false detections. |
点此下载全文
次