Improved K-means Clustering-based Defect Detection Method for Photovoltaic Panels
-
摘要: 为了能够对光伏组件热斑部分准确地识别和提取，提出了一种基于HSV空间模型的改进K均值聚类图像处理方法。首先，将红外图像进行HSV空间转换和双边滤波处理，去除噪声并提高图像对比度；其次，使用高斯核函数估计实现图像灰度概率密度函数提取，并以此获取初始聚类中心；最后，利用先验知识对图像进行K均值聚类，提取和量化热斑缺陷。研究结果表明，该方法能够快速地检测定位热斑位置并统计出光伏板损坏程度，具有较高的精度以及较好的灵敏性和稳定性。Abstract: An image processing method based on the HSV space model with an improved K-means clustering algorithm is proposed to accurately identify and extract the hot spot part of photovoltaic modules. First, the infrared image is transformed into the HSV space and bilaterally filtered to remove noise and improve the image contrast. Second, the Gaussian kernel function is used to extract the image grayscale probability density function, and then the initial clustering center is obtained. Finally, K-means clustering is applied to the image using prior knowledge to extract and quantify the hot spot defects. The research results show that the method can quickly detect and locate the hotspot position and calculate the degree of damage to the photovoltaic panel, and has high accuracy, good sensitivity, and stability.
-
Key words:
- infrared image /
- defect detection /
- hot spot /
- photovoltaic module /
- HSV space model /
- improved K-means clustering
-
图 2 图像预处理效果对比：(a) 原图；(b) 预处理后图像；(c) 原图灰度分布；(d) 预处理后图像灰度分布
Figure 2. Comparison of image pre-processing effects: (a) Original drawing; (b) Pre processed image; (c) Gray distribution of original image; (d) Gray distribution of preprocessing image
图 3 高斯核函数估计前后的灰度概率密度
Figure 3. Probability density of grayscale before and after estimation of Gaussian kernel function
图 10 不同算法缺陷检测对比：(a) 改进K均值算法；(b)传统K均值算法；(c) 改进OSTU算法；(d) B样条最小二乘拟合法
Figure 10. Comparison chart of different algorithms for defect detection; (a) Improved K-means algorithm; (b) Traditional K-means algorithm; (c) Improved OSTU algorithm; (d) B-spline least square fitting method
表 1 禅思H20T红外相机参数
Table 1. Zenmuse H20T infrared camera parameters
Resolution Wavelength range Measuring range Focal length 640×512 8-14 μm −40℃ to 150℃ 13.5 mm 
下载: 导出CSV
表 2 K-means改进效果评估
Table 2. Evaluation table of K-means improvement effect
Algorithm IE DBI SC Improved K-means 3.4843 0.4709 0.8274 Traditional K-means 3.5804 0.4719 0.8248
下载: 导出CSV
-
[1] Romero-Cadaval Enrique, Spagnuolo Giovanni, Garcia Franquelo Leopoldo, et al. Grid-connected photovoltaic generation plants: components and operation[J]. Industrial Electronics Magazine IEEE, 2013, 7(3): 6-20. doi: 10.1109/MIE.2013.2264540 [2] Vergura Silvano, Marino Francescomaria. Quantitative and computer-aided thermography-based diagnostics for PV devices: part Ⅰ—framework[J]. IEEE Journal of Photovoltaics, 2017, 7(3): 822-827. doi: 10.1109/JPHOTOV.2017.2655484 [3] YANG Ruizhen, DU Bolun, DUAN Puhong, et al. Electromagnetic induction heating and image fusion of silicon photovoltaic cell electrothermography and electroluminescence[J]. IEEE Transactions on Industrial Informatics, 2020, 16(7): 4413-4422. doi: 10.1109/TII.2019.2922680 [4] 卜迟武, 刘涛, 李锐, 等. 光伏电池缺陷红外热成像检测与图像序列处理[J]. 光学学报, 2022, 42(7): 118-124. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB202207012.htmBU Chiwu, LIU Tao, LI Rui, et al. Infrared thermal imaging detection and image sequence processing of photovoltaic cell defects[J]. Journal of Optics, 2022, 42(7): 118-124. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB202207012.htm [5] WANG Youyang, LI Liying, SUN Yifan, et al. Adaptive automatic solar cell defect detection and classification based on absolute electroluminescence imaging[J]. Energy, 2021, 229: 120606. doi: 10.1016/j.energy.2021.120606 [6] 韩金辉, 魏艳涛, 彭真明, 等. 红外弱小目标检测方法综述[J]. 红外与激光工程, 2022, 51(4): 438-461. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ202204050.htmHAN J H, WEI Y T, PENG Z M, et al. Overview of infrared dim target detection methods[J]. Infrared and Laser Engineering, 2022, 51(4): 438-461. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ202204050.htm [7] 夏杰锋, 唐武勤, 杨强. 光伏航拍红外图像的热斑自动检测方法[J]. 浙江大学学报(工学版), 2022, 56(8): 1640-1647. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDZC202208018.htmXIA J F, TANG W Q, YANG Q. Automatic detection method of hot spot in photovoltaic aerial infrared image[J]. Journal of Zhejiang University (Engineering Edition), 2022, 56(8): 1640-1647. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDZC202208018.htm [8] 蒋琳, 苏建徽, 施永, 等. 基于红外热图像处理的光伏阵列热斑检测方法[J]. 太阳能学报, 2020, 41(8): 180-184. https://www.cnki.com.cn/Article/CJFDTOTAL-TYLX202201055.htmJIANG L, SU J H, SHI Y, et al. Thermal spot detection method of photovoltaic array based on infrared thermal image processing[J]. Journal of Solar Energy, 2020, 41(8): 180-184. https://www.cnki.com.cn/Article/CJFDTOTAL-TYLX202201055.htm [9] NIE Junfei, LUO Ting, LI Hui. Automatic hotspots detection based on UAV infrared images for large-scale PV plant[J]. Electronics Letters, 2020, 56(19): 993-995. doi: 10.1049/el.2020.1542 [10] CHEN Haiyong, ZHAO Huifang, HAN Da, et al. Accurate and robust crack detection using steerable evidence filtering in electroluminescence images of solar cells[J]. Optics and Lasers in Engineering, 2019, 118(7): 22-33. [11] 钱晓亮, 张鹤庆, 张焕龙, 等. 基于视觉显著性的太阳能电池片表面缺陷检测[J]. 仪器仪表学报, 2017, 38(7): 1570-1578. https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201707002.htmQIAN X L, ZHANG H Q, ZHANG H L, et al. Visual saliency-based surface defect detection of solar cells[J]. Journal of Instrumentation, 2017, 38(7): 1570-1578. https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201707002.htm [12] 王培珍, 郑诗程. 基于红外图像的太阳能光伏阵列故障分析[J]. 太阳能学报, 2010, 31(2): 197-202. https://www.cnki.com.cn/Article/CJFDTOTAL-TYLX201002013.htmWANG P Z, ZHENG S C. Fault analysis of solar photovoltaic arrays based on infrared images[J]. Journal of Solar Energy, 2010, 31(2): 197-202. https://www.cnki.com.cn/Article/CJFDTOTAL-TYLX201002013.htm [13] 赵玲玲, 汪烨, 刘俊. 基于无人机与HSV空间的光伏电池板检测分析[J]. 红外技术, 2020, 42(10): 978-982. http://hwjs.nvir.cn/article/id/hwjs202010010ZHAO Lingling, WANG Ye, LIU Jun. Analysis of PV panel inspection based on UAV and HSV space[J]. Infrared Technology, 2020, 42(10): 978-982. http://hwjs.nvir.cn/article/id/hwjs202010010 [14] 李茂月, 刘泽隆, 赵伟翔, 等. 面结构光在机检测的叶片反光抑制技术[J]. 中国光学, 2022, 15(3): 464-475. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGA202203007.htmLI M Y, LIU Z L, ZHAO W X, et al. Leaf reflection suppression technique for in-camera detection of surface structured light[J]. China Optics, 2022, 15(3): 464-475. https://www.cnki.com.cn/Article/CJFDTOTAL-ZGGA202203007.htm [15] 姜斌, 石峰, 崔东旭, 等. 改进的K均值聚类红外目标检测方法[J]. 应用光学, 2012, 33(4): 766-769. https://www.cnki.com.cn/Article/CJFDTOTAL-YYGX201204025.htmJIANG B, SHI F, CUI D X, et al. Improved k-means clustering infrared target detection method[J]. Applied Optics, 2012, 33(4): 766-769. https://www.cnki.com.cn/Article/CJFDTOTAL-YYGX201204025.htm [16] 王骏, 王士同, 邓赵红, 等. 面向小目标图像的快速核密度估计图像阈值分割算法[J]. 自动化学报, 2012, 38(10): 1679-1689. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO201210016.htmWANG J, WANG S T, DENG Z H, et al. Fast kernel density estimation image threshold segmentation algorithm for small target images[J]. Journal of Automation, 2012, 38(10): 1679-1689. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO201210016.htm [17] Janssen Paul, Swanepoel Jan. Efficiency behaviour of kernel-smoothed kernel distribution function estimators[J]. South African Statistical Journal, 2020, 54(1): 15-23. doi: 10.37920/sasj.2020.54.1.2 [18] HAN Bohyung, Comaniciu Dorin, ZHU Ying, et al. Sequential kernel density approximation and its application to real-time visual tracking[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2008, 30(7): 1186-97. [19] HAN Ping, WANG Wenqing, SHI Qingyan, et al. A combined online-learning model with K-means clustering and GRU neural networks for trajectory prediction[J]. Ad Hoc Networks, 2021, 117(44): 102476. [20] 廖纪勇, 吴晟, 刘爱莲. 基于相异性度量选取初始聚类中心改进的K-means聚类算法[J]. 控制与决策, 2021, 36(12): 3083-3090. https://www.cnki.com.cn/Article/CJFDTOTAL-KZYC202112028.htmLIAO J Y, WU S, LIU A L. Improved K-means clustering algorithm for selecting initial clustering centers based on the phase dissimilarity measure[J]. Control and Decision Making, 2021, 36(12): 3083-3090. https://www.cnki.com.cn/Article/CJFDTOTAL-KZYC202112028.htm [21] 王惠琴, 侯文斌, 彭清斌, 等. 基于K均值聚类的SPPM分步分类检测算法[J]. 通信学报, 2022, 43(1): 161-171. https://www.cnki.com.cn/Article/CJFDTOTAL-TXXB202201015.htmWANG H Q, HOU W B, PENG Q B, et al. SPPM stepwise classification and detection algorithm based on K-mean clustering[J]. Journal of Communication, 2022, 43(1): 161-171. https://www.cnki.com.cn/Article/CJFDTOTAL-TXXB202201015.htm [22] 徐黎明, 吕继东. 基于同态滤波和K均值聚类算法的杨梅图像分割[J]. 农业工程学报, 2015, 31(14): 202-208. https://www.cnki.com.cn/Article/CJFDTOTAL-NYGU201514028.htmXU L M, LV J D. Bayberry image segmentation based on homomorphic filtering and K-means clustering algorithm[J]. Journal of Agricultural Engineering, 2015, 31(14): 202-208. https://www.cnki.com.cn/Article/CJFDTOTAL-NYGU201514028.htm [23] 欧阳周, 张怀亮, 唐子暘, 等. 复杂纹理瓷砖表面缺陷检测算法研究[J]. 西北工业大学学报, 2022, 40(2): 414-421. https://www.cnki.com.cn/Article/CJFDTOTAL-XBGD202202023.htmOUYANG Z, ZHANG H L, TANG Z Y, et al. Research on surface defect detection algorithm for complex textured tiles[J]. Journal of Northwestern Polytechnical University, 2022, 40(2): 414-421. https://www.cnki.com.cn/Article/CJFDTOTAL-XBGD202202023.htm [24] 肖懿, 罗丹, 蒋沁知, 等. 基于温度概率密度的变电站高压设备故障热红外图像识别方法[J]. 高电压技术, 2022, 48(1): 307-318. https://www.cnki.com.cn/Article/CJFDTOTAL-GDYJ202201030.htmXIAO Y, LUO D, JIANG Q Z, et al. Thermal infrared image identification method for high-voltage equipment faults in substations based on temperature probability density[J]. High Voltage Technology, 2022, 48(1): 307-318. https://www.cnki.com.cn/Article/CJFDTOTAL-GDYJ202201030.htm [25] 杨磊, 高德东, 孟广双, 等. 光伏组件故障诊断方法研究[J]. 机械设计与制造, 2021(12): 82-87. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYZ202112018.htmYANG L, GAO D D, MENG G S, et al. Research on fault diagnosis method of photovoltaic module[J]. Mechanical Design and Manufacturing, 2021(12): 82-87. https://www.cnki.com.cn/Article/CJFDTOTAL-JSYZ202112018.htm [26] Vega Diaz, Vlaminck Michiel, Lefkaditis Dionysios, et al. Solar panel detection within complex backgrounds using thermal images acquired by UAVs[J]. Sensors (Basel), 2020, 20(21): 6219. [27] 李喆雨, 丁坤, 张经炜, 等. 基于选择增强的光伏阵列遮挡分割研究[J]. 激光与光电子学进展, 2021, 58(24): 201-209. https://www.cnki.com.cn/Article/CJFDTOTAL-JGDJ202124017.htmLI Z Y, DING K, ZHANG J W, et al. Selective enhancement-based shading segmentation of photovoltaic arrays[J]. Advances in Laser and Optoelectronics, 2021, 58(24): 201-209. https://www.cnki.com.cn/Article/CJFDTOTAL-JGDJ202124017.htm -
点击查看大图
图(10) / 表(3)
计量
- 文章访问数: 29
- HTML全文浏览量: 10
- PDF下载量: 4
- 被引次数: 0
