| 基于改进U-net和CNN的绝缘子自爆检测方法研究 |
| |
| 引用本文： | 李俊,任景,王晔琳,张小东,薛晨,任冲,范国伟.基于改进U-net和CNN的绝缘子自爆检测方法研究[J].陕西电力,2021,0(8):98-103. |
| |
| 作者姓名： | 李俊 任景 王晔琳 张小东 薛晨 任冲 范国伟 |
| |
| 作者单位： | （1. 国家电网公司西北分部，陕西 西安 710048； 2. 西安理工大学 电气学院，陕西 西安 710048） |
| |
| 摘 要： | 针对绝缘子自爆故障人工检测效率低，成本高的问题，基于改进U-net和卷积神经网络（CNN）模型，提出一种可有效识别绝缘子自爆故障的双阶段目标检测算法。首先，在语义分割阶段使用改进U-net模型，通过翻倍提高图像分辨率的方法有效提高图像分割精度。其次，在图像分类阶段提出更适合所提问题且有效提高分类准确度的新型CNN模型。最后，使用无人机拍摄的绝缘子图片为实验数据进行实验。实验结果表明所提算法识别精度较高。
|
| 关 键 词： | 绝缘子 自爆故障 改进U-net 卷积神经网络 双阶段目标检测算法 |
Insulator Self-explosion Detection Method Based on Improved U-net and CNN |
| |
| LI Jun,REN Jing,WANG Yelin,ZHANG Xiaodong,XUE Chen,REN Chong,FAN Guowei.Insulator Self-explosion Detection Method Based on Improved U-net and CNN[J].Shanxi Electric Power,2021,0(8):98-103. |
| |
| Authors: | LI Jun REN Jing WANG Yelin ZHANG Xiaodong XUE Chen REN Chong FAN Guowei |
| |
| Affiliation: | （1. Northwest Branch of State Grid Corporation of China，Xi’an 710048，China；2. School of Electrical Engineering，Xi’an University of Technology，Xi’an 710048，China） |
| |
| Abstract: | Aiming at the low efficiency and high cost of manual detection of insulator self-explosion fault，two-stage target detection algorithm based on improved U-net and convolutional neural network（CNN） model is proposed，which can effectively identify insulator self explosion fault. Firstly，the improved U-net model is used in semantic segmentation stage， and the image segmentation accuracy is improved effectively by doubling the image resolution. Secondly，a new CNN model which is more suitable for this problem and can effectively improve the classification accuracy is proposed in the image classification stage. Finally，the insulator image taken by UAV is used as the experimental data. Experimental results show that the algorithm can effectively identify insulator defects. |
| |
| Keywords: | insulator self-explosion fault improved U-net convolution neural network two-stage target detection algorithm |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《陕西电力》浏览原始摘要信息 |
|
点击此处可从《陕西电力》下载全文 |
