基于 Pix2PixHD 和 Star⁃CAA⁃YOLOv8 的双端故障行波定位方法

打开文本图片集

关键词：双端行波；故障定位；小波变换；Pix2PixHD；YOLOv8；StarNet；CAA；目标检测中图分类号：TN713⁃34；TM98 文献标识码： A 文章编号：1004⁃373X（2025）12⁃0157⁃10

Abstract：In allusion to the traditional fault traveling wave method applied in transmission lines， which is subject to the problem of low sampling frequency of sampling equipment and low localization accuracy， a double ⁃ ended traveling wave fault localization algorithm combining antagonistic generative network and neural network is proposed. In the method， the wavelet transform is used to extract the initial wavehead time ⁃ domain features of the faulty traveling wave at low sampling rates， the Pix2PixHD is input in the form of pictures， and the obtained initial traveling wave time⁃domain features at high sampling rates is output， thereby improving the fault localization accuracy. A Star⁃CAA⁃YOLOv8 model is proposed for traveling wave detection to evaluate the quality of the generated images， and the instance segmentation function is used to realize the automated reading of traveling wave arrival time to solve the inconvenient reading problem brought by the images. The fault traveling wave image at 1 MHz sampling rate is output by means of Pix2PixHD to obtain the fault traveling wave image at 2 MHz sampling rate. The absolute error of fault location is reduced by 83.33% and the relative error is reduced by 84.44% . The method has the possibility of narrowing down lower errors. The multiple simulation and comparison experiments show that the proposed method can accurately obtain fault data at high sampling rates and has good generalization and practicality.

Keywords：double⁃ended traveling wave; fault localization; wavelet transform; Pix2PixHD; YOLOv8; StarNet; CAA; object detection

0 引 言

配电网需快速、准确定位线路故障，迅速隔离故障区段并实施抢修，这是保障电力系统安全稳定运行的关键[1]。（剩余10214字）