基于梯度图像融合的接触网绝缘子故障检测

石杰; 张靖; 钟汉华

基于梯度图像融合的接触网绝缘子故障检测

石杰^1,,
张靖^2, ,,
钟汉华¹

1.
华东交通大学电气与自动化工程学院, 江西南昌 330000
2.
昭通学院物理与信息工程学院, 云南昭通 657000

基金项目:

江西省教育厅青年基金项目 GJJ2200659

国家自然科学基金 52267015

详细信息

作者简介:
石杰（1990-），男，硕士，研究方向：电力系统故障分析与诊断。E-mail：1003569980@qq.com

通讯作者:
张靖（1994-），男，硕士，研究方向：电气设备故障诊断，E-mail：1780040544@qq.com

中图分类号: TM854, TP391.4
计量
- 文章访问数: 35
- HTML全文浏览量: 9
- PDF下载量: 15
- 被引次数: 0
出版历程
- 收稿日期: 2022-11-27
- 修回日期: 2023-01-19
- 刊出日期: 2023-10-20

Catenary Insulator Fault Detection Based on Gradient Image Fusion

SHI Jie^1
,,
ZHANG Jing^{2
, ,},
ZHONG Hanhua¹

1.
School of Electrical and Automation Engineering, East China Jiaotong University, Nanchang 330000, China
2.
School of Physics and Information Engineering of Zhaotong University, Zhaotong 657000, China

摘要

摘要: 针对单一红外图像或可见光图像不能够实现全天候检测的问题，提出了一种梯度图像融合模型将红外和可见光图像进行融合。先采用加速稳健特征算法（speeded-up robust features，SURF）将两幅图像的特征点进行匹配。接着采样剪切波变换（non-subsampled shearlet transform, NSST）算法将待融合图像进行分解，形成具有高频分量信息和低频分量信息的图，再分别对绝缘子的高频分量图和低频分量图进行融合，实现局部融合。利用NSST的逆变换对高频分量图和低频分量图进行逆变换，得到最终融合图，实现全局融合。对融合图像进行质量评价。采用最小二乘法直线拟合算法在二值图像的基础上来实现绝缘子的自爆检测；采用像素积分投影法来检测绝缘子片裂纹情况；采用颜色特征来检测绝缘子表面是否存在污秽的情况。通过实验对比单张图像和融合图像的检测结果的准确率。实验结果表明，采用基于融合图像的绝缘子自爆、绝缘子片裂纹、绝缘子表面污秽3个故障的识别率分别达到了95%、91%、90%，均高于单一的红外图像或可见光图像的识别率。
- 接触网绝缘子故障 /
- 图像融合 /
- 局部融合 /
- 全局融合 /
- 图像质量评价
Abstract: To accurately identify a single infrared or visible image under all weather conditions, a gradient image fusion model is proposed to fuse infrared and visible images. First, the accelerated up robust features algorithm is used to match the feature points of the two images. Further, the sampled shear wave transform (NSST) algorithm decomposes the image to be fused to form a map with high-frequency and low-frequency component information and then fuses the high-frequency and low-frequency component maps of insulators to achieve local fusion. The high- and low-frequency component maps are inversely transformed by the inverse transform of the NSST to obtain the final fusion map and achieve global fusion. The quality of the fused images is also evaluated. The line fitting algorithm based on the least-squares method is used to detect insulator self-explosions based on a binary image, the pixel integral projection method is used to detect cracks in the insulator, and color features are used to detect whether the insulator surface is polluted. The accuracies of the detection results of a single image and fusion image were compared through experiments. The experimental results show that the recognition rates of insulator self-explosion, insulator cracks, and insulator surface contamination based on fusion images are 95%, 91%, and 90%, respectively, which are higher than the recognition rates of single infrared image or visible image.
- OCS insulator fault /
- image fusion /
- local fusion /
- global integration /
- image quality evaluation

HTML全文

图 1 图像配准流程

Figure 1. Image registration process

下载: 全尺寸图片幻灯片

图 2 图像配准实验效果

Figure 2. Experimental results of image registration

下载: 全尺寸图片幻灯片

图 3 图像梯度融合流程

Figure 3. Image gradient fusion process

下载: 全尺寸图片幻灯片

图 4 NSST图像分解

Figure 4. NSST image decomposition

下载: 全尺寸图片幻灯片

图 5 PCNN网络模型

Figure 5. PCNN network model

下载: 全尺寸图片幻灯片

图 6 红外和可见光图像融合的图像

Figure 6. Fusion of infrared and visible light images

下载: 全尺寸图片幻灯片

图 7 不同方法下的图像融合

Figure 7. Image fusion under different methods

下载: 全尺寸图片幻灯片

图 8 融合NSCT和二维最大熵分割方法分割效果

Figure 8. Fusion of NSCT and two-dimensional maximum entropy segmentation method results

下载: 全尺寸图片幻灯片

图 9 完好绝缘子像素积分投影（左）和裂纹绝缘子像素积分投影（右）

Figure 9. Pixel integral projection of intact insulator (left) and cracked insulator (right)

下载: 全尺寸图片幻灯片

表 1 不同评价指标的得分和排名

Table 1. Scores and rankings of different evaluation indicators

Threshold t	Q^AB/F	EN	MI	MSF	MSE	SSIM	Weighted ranking
0.3	0.617	6.654	1.299	4.193	20.176	0.608	4
0.4	0.273	6.78	1.68	3.212	13.882	0.608	6
0.5	0.639	6.596	2.512	4.155	20.389	0.745	3
0.6	0.657	6.714	3.669	7.26	23.64	0.801	1
0.7	0.656	6.691	2.766	6.458	20.29	0.699	2
0.8	0.645	6.674	3.663	5.966	13.59	0.802	5

下载: 导出CSV

表 2 正常绝缘子在不同融合方法下的图像质量评价指标

Table 2. Image quality evaluation indexes of normal insulators under different fusion methods

Fusion method	Q^AB/F	EN	MI	MSF	MSE	SSIM
LAB	0.41148	5.78892	7.36956	4.96356	39.2316	0.8004
IHS	0.37872	5.24208	6.88776	6.24396	42.6876	0.7326
Weighted average method	0.41268	5.1318	7.6392	8.1084	46.4256	0.85464
Brovey	0.3774	6.44232	8.96232	6.16032	39.8748	0.67764
Wavelet transform	0.44604	7.6032	8.05068	6.12636	39.5676	0.44376
PCA	0.71208	6.69408	9.24612	7.58172	47.4444	1.04532
Gradient image fusion model	0.83712	7.2426	9.49392	8.84436	51.2112	1.07448

下载: 导出CSV

表 3 绝缘子自爆情况下不同融合方法下的图像质量评价指标

Table 3. Image quality evaluation indexes under different fusion methods in the case of insulator self-explosion

Fusion method	Q^AB/F	EN	MI	MSF	MSE	SSIM
LAB	0.15719	5.08651	6.53543	4.32993	35.7423	0.5137
IHS	0.12716	4.58524	6.09378	5.50363	38.9103	0.45155
Weighted average method	0.15829	4.48415	6.7826	7.2127	42.3368	0.56342
Brovey	0.12595	5.68546	7.99546	5.42696	36.3319	0.40117
Wavelet transform	0.18887	6.7496	7.15979	5.39583	36.0503	0.18678
PCA	0.43274	5.91624	8.25561	6.72991	43.2707	0.76494
Gradient image fusion model	0.54736	6.41905	8.48276	7.88733	46.7236	0.73821

下载: 导出CSV

表 4 绝缘子裂纹情况下裂纹在不同融合方法下的图像质量评价指标

Table 4. Image quality evaluation index of crack under different fusion methods under the condition of insulator crack

Fusion method	Q^AB/F	EN	MI	MSF	MSE	SSIM
LAB	0.17290	5.595161	7.188973	4.762923	39.31653	0.56507
IHS	0.13987	5.043764	6.703158	6.053993	42.80133	0.496705
Weighted average method	0.17411	4.932565	7.46086	7.93397	46.57048	0.619762
Brovey	0.13854	6.254006	8.795006	5.969656	39.96509	0.441287
Wavelet transform	0.20775	7.42456	7.875769	5.935413	39.65533	0.205458
PCA	0.47601	6.507864	9.081171	7.402901	47.59777	0.841434
Gradient image fusion model	0.60209	7.060955	9.331036	8.676063	51.39596	0.812031

下载: 导出CSV

表 5 绝缘子污秽在不同融合方法下的图像质量评价指标

Table 5. Image quality evaluation index of insulator pollution under different fusion methods

Fusion method	Q^AB/F	EN	MI	MSF	MSE	SSIM
LAB	0.26719	5.19651	6.64543	4.43993	35.8523	0.6237
IHS	0.23716	4.69524	6.20378	5.61363	39.0203	0.56155
Weighted average method	0.26829	4.59415	6.8926	7.3227	42.4468	0.67342
Brovey	0.23595	5.79546	8.10546	5.53696	36.4419	0.51117
Wavelet transform	0.29887	6.8596	7.26979	5.50583	36.1603	0.29678
PCA	0.54274	6.02624	8.36561	6.83991	43.3807	0.87494
Gradient image fusion model	0.65736	6.52905	8.59276	7.99733	46.8336	0.84821

下载: 导出CSV

表 6 绝缘子故障检测的准确率对比

Table 6. Comparison of the accuracy of insulator fault detection

Image type recognition accuracy	An insulator without defects	The exploding insulator	Cracked insulator	The contaminated insulator
Infrared image	0.90	0.93	0.88	0.87
Visible light image	0.93	0.9	0.91	0.89
Fused image	0.94	0.95	0.91	0.90

下载: 导出CSV

参考文献(20)

[1]	顾桂梅, 陈国翠. 改进GA-BP算法的棒式绝缘子表面缺陷识别[J]. 铁道科学与工程学报, 2022, 19(2): 546-553. https://www.cnki.com.cn/Article/CJFDTOTAL-CSTD202202029.htm GU Guimei, CHEN Guocui. Surface defect identification of rod insulator with improved GA-BP algorithm[J]. Journal of Railway Science and Engineering, 2022, 19(2): 546-553. https://www.cnki.com.cn/Article/CJFDTOTAL-CSTD202202029.htm
[2]	谭平, 李旭峰, 丁进, 等. 基于Mask R-CNN和多特征聚类模型的接触网绝缘子识别和缺陷检测(英文)[J]. 浙江大学学报-理学 A(应用物理与工程) , 2022, 23(9): 745-757. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDYG200702003.htm TAN Ping, LI Xufeng, DING Jin, et al. Insulator identification and defect detection of catenary based on Mask R-CNN and multi-feature clustering model[J]. Journal of Zhejiang University-Science A(Applied Physics and Engineering), 2022, 23(9): 745-757. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDYG200702003.htm
[3]	刘新伯, 李昊, 陈强, 等. 复杂背景下多个绝缘子串红外图像分割方法[J]. 计算技术与自动化, 2022, 41(2): 125-130. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJH202202023.htm LIU Xinbo, LI Hao, CHEN Qiang, et al. Infrared image segmentation of multiple insulator strings in complex background[J]. Computational Technology & Automation, 2022, 41(2): 125-130. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJH202202023.htm
[4]	董懿飞, 舒胜文, 陈诚, 等. 基于红外图像分割与SSA-SVM的复合绝缘子缺陷检测方法[J]. 电气技术, 2021, 22(11): 73-79. https://www.cnki.com.cn/Article/CJFDTOTAL-DQJS202111014.htm DONG Yifei, SHU Shengwen, CHEN Cheng, et al. Composite insulator defect detection method based on infrared image segmentation and SSA-SVM [J]. Electrical Technology, 2021, 22(11): 73-79. https://www.cnki.com.cn/Article/CJFDTOTAL-DQJS202111014.htm
[5]	朱文天. 输电线路绝缘子识别与光谱缺陷检测技术研究[D]. 南宁: 广西大学, 2019. ZHU W T. Research on Insulator Identification and Spectral Defect Detection Technology of Transmission Line[D]. Nanning: Guangxi University, 2019.
[6]	高世闯. 基于全卷积神经网络的绝缘子识别与故障检测[D]. 石家庄: 石家庄铁道大学, 2020. GAO Shichuang. Insulator Identification and Fault Detection Based on Convolutional Neural Network[D]. Shijiazhuang: Shijiazhuang Railway University, 2020.
[7]	汝承印, 张仕海, 张子淼, 等. 基于轻量级MobileNet-SSD和MobileNetV2-DeeplabV3+的绝缘子故障识别方法[J]. 高电压技术, 2022, 48(9): 3670-3679. https://www.cnki.com.cn/Article/CJFDTOTAL-GDYJ202209032.htm RU Chengyin, ZHANG Shihai, ZHANG Zimiao, et al. Insulator fault identification method based on lightweight MobileNet-SSD and MobileNetV2-DeeplabV3+[J]. High Voltage Technology, 2022, 48(9): 3670-3679. https://www.cnki.com.cn/Article/CJFDTOTAL-GDYJ202209032.htm
[8]	马巧慧. 基于多特征融合的绝缘子定位与故障识别的研究[D]. 大庆: 东北石油大学, 2021. MA Qiaohui. Research on Insulator Location and Fault Identification Based on Multi-feature Fusion[D]. Daqing: Northeast Petroleum University, 2021.
[9]	周亮君, 肖世德, 李晟尧, 等. 基于SURF与GPU加速数字图像处理[J]. 传感器与微系统, 2022, 41(3): 98-100. https://www.cnki.com.cn/Article/CJFDTOTAL-CGQJ202203026.htm ZHOU Liangjun, XIAO Shide, LI Shengyao, et al. Digital image processing accelerated by SURF and GPU[J]. Sensors and Microsystems, 2022, 41(3): 98-100. https://www.cnki.com.cn/Article/CJFDTOTAL-CGQJ202203026.htm
[10]	陈熹, 张娜娜. 基于Hessian矩阵的车道线检测方法[J]. 电视技术, 2022, 46(2): 168-172. https://www.cnki.com.cn/Article/CJFDTOTAL-DSSS202202040.htm CHEN Xi, ZHANG Nana. Lane line detection method based on hessian matrix [J]. Television Technology, 2022, 46(2): 168-172. https://www.cnki.com.cn/Article/CJFDTOTAL-DSSS202202040.htm
[11]	余波, 吴静, 周琦宾. 一种基于改进Canny算子的边缘检测算法[J]. 制造业自动化, 2022, 44(8): 24-26, 43. https://www.cnki.com.cn/Article/CJFDTOTAL-JXGY202208006.htm YU Bo, WU Jing, ZHOU Qibin. An edge detection algorithm based on improved Canny operator[J]. Manufacturing Automation, 2022, 44(8): 24-26, 43. https://www.cnki.com.cn/Article/CJFDTOTAL-JXGY202208006.htm
[12]	崔晓荣, 沈涛, 黄建鲁, 等. FABEMD和改进局部能量窗口的红外中波和长波图像融合[J]. 光谱学与光谱分析, 2021, 41(7): 2043-2049. https://www.cnki.com.cn/Article/CJFDTOTAL-GUAN202107009.htm CUI Xiaorong, SHEN Tao, HUANG Jianlu, et al. FABEMD and improved local energy window for infrared medium and long wave image fusion[J]. Spectroscopy and Spectral Analysis, 2021, 41(7): 2043-2049. https://www.cnki.com.cn/Article/CJFDTOTAL-GUAN202107009.htm
[13]	杨晓城, 杨真乙, 阎敬业, 等. 基于拉普拉斯算子的综合孔径辐射计成像算法[J]. 电子学报, 2022, 50(2): 339-345. https://www.cnki.com.cn/Article/CJFDTOTAL-DZXU202202009.htm YANG Xiaocheng, YANG Zhenyi, YAN Jingye, et al. Integrated aperture radiometer imaging algorithm based on Laplace operator[J]. Journal of Electronics, 2022, 50(2): 339-345. https://www.cnki.com.cn/Article/CJFDTOTAL-DZXU202202009.htm
[14]	吴肖楚. 基于非下采样剪切波变换的低照度图像增强算法研究[D]. 南京: 南京邮电大学, 2020. WU Xiaochu. Research on Low Illuminance Image Enhancement Algorithm Based on Non Downsampling Shear Wave Transform[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2020.
[15]	石森, 李小平, 徐家梁. 基于能量加权溶解气体分析的IPSO-BP变压器早期故障诊断[J]. 制造业自动化, 2022, 44(5): 43-45, 50. https://www.cnki.com.cn/Article/CJFDTOTAL-JXGY202205011.htm SHI Sen, LI Xiaoping, XU Jialiang. Early fault diagnosis of IPSO-BP transformer based on energy weighted dissolved gas analysis[J]. Manufacturing Automation, 2022, 44(5): 43-45, 50. https://www.cnki.com.cn/Article/CJFDTOTAL-JXGY202205011.htm
[16]	张捍卫, 李晓玲, 杨永勤, 等. 拉普拉斯算子[J/OL] [2022-10-13]. 导航定位学报: 1-8. https://www.doc88.com/p-39339696421316.html. ZHANG Hanwei, LI Xiaoling, YANG Yongqin, et al. Laplace operator [J/OL] [2022-10-13]. Journal of Navigation and Positioning: 1-8. https://www.doc88.com/p-39339696421316.html.
[17]	刘子英, 张靖, 邓芳明. 基于BP神经网络的高压隔离开关分合闸监测识别[J]. 电力系统保护与控制, 2020, 48(5): 134-140. https://www.cnki.com.cn/Article/CJFDTOTAL-JDQW202005018.htm LIU Ziying, ZHANG Jing, DENG Fangming. Monitoring and identification of high voltage isolation switch opening and closing based on BP neural network[J]. Power System Protection and Control, 2020, 48(5): 134-140. https://www.cnki.com.cn/Article/CJFDTOTAL-JDQW202005018.htm
[18]	York D. Least squares fitting of a straight line with correlated errors[J]. Earth and Planetary Science Letters, 1968, 5: 320-324.
[19]	崔庆华, 程科, 李肇基. 积分投影与连通域法结合的人眼定位[J]. 计算机与数字工程, 2019, 47(4): 949-953. https://www.cnki.com.cn/Article/CJFDTOTAL-JSSG201904043.htm CUI Qinghua, CHENG Ke, LI Zhaoji. The combination of integral projection and connected domain method for human eye localization[J]. Computer and Digital Engineering, 2019, 47(4): 949-953. https://www.cnki.com.cn/Article/CJFDTOTAL-JSSG201904043.htm
[20]	赵文竹. 基于图像颜色特征提取的维生素C含量测定方法研究[D]. 淮北: 淮北师范大学, 2021. ZHAO Wenzhu. Research on the Method for Determining Vitamin C Content Based on Image Color Feature Extraction[D]. Huaibei: Huaibei Normal University, 2021.