Detection method of a lead seal crack in a high-voltage cable based on laser atlas

Current detection methods for lead seal cracks in high-voltage cables have low accuracy in detecting small or shallow cracks, which can easily be missed or misjudged, thereby affecting the detection effect. Therefore, a laser image-based method for detecting lead seal cracks in high-voltage cables was proposed. First, a device for collecting the laser spectrum of high-voltage cable lead seals is designed, and the laser spectrum of high-voltage cable lead seals is collected using unmanned aerial vehicles. Then, based on the collected laser spectrum of high-voltage cable lead seals, a wavelet packet energy feature extraction method and bispectral estimation method are used to extract the energy characteristics, depth, and angle detection parameters of high-voltage cable lead seal crack defects. Finally, a dynamic generalized regression network model is constructed, which considers the energy characteristics, depth, and angle detection parameters of high-voltage cable lead seal crack defects as inputs to achieve high-voltage cable lead seal crack defect detection. The experiment shows that this method can effectively extract the energy characteristics of the laser spectrum signal of high-voltage cable lead seals and accurately monitor the number of cracked lead seals and lead seals with cracks exceeding 5 cm on the high-voltage cable line. This can effectively provide technical support for the maintenance of high-voltage cables.