Hyperspectral-Based Corona Aging Evaluation for Composite Insulators
-
摘要: 复合绝缘子由于其良好的憎水性和憎水迁移性在输电线路上得到了广泛应用,而电晕放电会造成复合绝缘子老化加剧而丧失性能. 为此提出了一种基于高光谱技术的复合绝缘子电晕老化状态评估方法. 首先,对全新硅橡胶复合绝缘片进行电晕老化,分析样本的傅里叶红外光谱变化,以傅里叶红外光谱图像作为老化状态分类的依据,将样品分为6个类别;然后,利用高光谱成像仪获取硅橡胶片表面不同波段的反射强度,采用主成分分析(principal component analysis,PCA)对原始谱线进行特征提取;最后,建立基于支持向量机的电晕老化状态评估(support vector machines-insulator corona aging status evaluation,SVM-CASE)模型,对60组预测数据进行分类验证,并对比分析了不同核函数对于模型评估准确率的影响. 高光谱检测及评估结果表明:不同老化时间下试样的高光谱图像有明显的区别,随着老化时间的增加,硅橡胶绝缘材料的光谱曲线在600~900 nm呈现反射率下降趋势;采用PCA算法进行特征提取后,利用polynomial核函数建立的评估模型的分类准确率达93.333%.
-
关键词:
- 高光谱技术 /
- 电晕老化 /
- 傅里叶红外光谱(FTIR) /
- 支持向量机 /
- 主成分分析
Abstract: Composite insulators have been widely used in transmission lines due to good water repellency and hydrophobic migration. Corona discharges aggravate the aging of composite insulators. To this end, a method was proposed to evaluate the corona aging of composite insulators based on hyperspectral technology. Firstly, the new composite insulating silicone rubber sheet was subjected to corona aging, the Fourier infrared spectrum of the samples was analyzed, by which the samples were classified into six categories. Then, the reflection intensity at different bands on the surface of the silicone rubber sheet was obtained by using a hyperspectral imaging camera. Further, the characteristics of the original spectra lines was extracted by the principal component analysis (PCA). Finally, the support vector machines-insulator corona aging evaluation (SVM-CAE) model was established, and 60 sets of data were examined. The effects of different kernel functions on the accuracy of model evaluation were compared. The results of hyperspectral detection and evaluation show that, as the aging time increases, the high spectrum of samples under different aging times has obvious difference. With the increase of aging time, the spectral curves of silicone rubber insulation show a decreasing trend in 600 nm−900 nm bands. After the feature extraction by PCA algorithm, the model classification accuracy with polynomial kernel function was 93.333%. -
表 1 前6主成分贡献率之和
Table 1. Sum of top 6 principal component contributions
主成分 PC1 PC2 PC3 PC4 PC5 PC6 贡献率/% 89.84 93.99 94.73 94.90 94.99 95.03 表 2 不同核函数下测试集数据的分类准确率
Table 2. Classification accuracy of test set data under different kernel functions
核函数类型 准确率/% SVM网络参数 linear (55/60) 91.667 -c 8 -g 32 -t 0 polynomial (56/60) 93.333 -c 8 -g 32 -t 1 radial basis of function (52/60) 86.667 -c 8 -g 32 -t 2 注:括号内数据表示正确组数/预测总组数. -
巢亚锋,高俊伟,岳一石,等. 湖南电网复合绝缘子带电自然积污特性[J]. 高电压技术,2018,44(9): 2799-2806.CHAO Yafeng, GAO Junwei, YUE Yishi, et al. Natural pollution characteristics of composite insulators for overhead transmission lines in hunan power grid[J]. High Voltage Engineering, 2018, 44(9): 2799-2806. 陈奇. 高压复合绝缘子应用及老化状态研究综述[J]. 绝缘材料,2016,49(4): 7-13,18.CHEN Qi. Summary of research on application and aging state of high voltage composite insulators[J]. Insulating Materials, 2016, 49(4): 7-13,18. 黄成才,李永刚. 复合绝缘子老化状态评估方法研究综述[J]. 电力建设,2014,35(9): 28-34. doi: 10.3969/j.issn.1000-7229.2014.09.006HUANG Chengcai. A review of research on evaluation methods of aging state of composite insulators[J]. Electric Power Construction, 2014, 35(9): 28-34. doi: 10.3969/j.issn.1000-7229.2014.09.006 张冠军,赵林,周润东,等. 硅橡胶复合绝缘子老化表征评估研究的现状与进展[J]. 高压电器,2016,52(4): 1-15.ZHANG Guanjun, ZHAO Lin, ZHOU Rundong, et al. Current status and progress of research on aging characterization evaluation of silicone rubber composite insulators[J]. High Voltage Apparatus, 2016, 52(4): 1-15. 谭宇翔,任海燕. 基于傅立叶红外光谱法的复合绝缘子老化程度诊断研究[J]. 电气自动化,2013,35(1): 91-94. doi: 10.3969/jissn.1000-3886.2013.01.033TAN Yuxiang, REN Haiyan. Diagnosis of aging degree of composite insulators based on Fourier transform infrared spectroscopy[J]. Electrical Automation, 2013, 35(1): 91-94. doi: 10.3969/jissn.1000-3886.2013.01.033 孙俊,金夏明,毛罕平,等. 高光谱图像技术在掺假大米检测中的应用[J]. 农业工程学报,2014,30(21): 301-307. doi: 10.3969/j.issn.1002-6819.2014.21.036SUN Jun, JIN Xiaming, MAO Hanping, et al. Application of hyperspectral image technology in detection of adulterated rice[J]. Transactions of the Chinese Society of Agricultural Engineering, 2014, 30(21): 301-307. doi: 10.3969/j.issn.1002-6819.2014.21.036 谢安国. 冷冻冷藏过程中猪肉的光谱特性研究及其品质的快速检测[D]. 广州: 华南理工大学, 2016. 杨小玲. 高光谱图像技术检测玉米种子品质研究[D]. 杭州: 浙江大学, 2016. 郭文川,董金磊. 高光谱成像结合人工神经网络无损检测桃的硬度[J]. 光学精密工程,2015,23(6): 1530-1537.GUO Wenchuan, DONG Jinlei. Non-destructive detection of peach hardness by hyperspectral imaging combined with artificial neural network[J]. Optics and Precision Engineering, 2015, 23(6): 1530-1537. 马本学,应义斌,饶秀勤,等. 高光谱成像在水果内部品质无损检测中的研究进展[J]. 光谱学与光谱分析,2009,29(6): 1611-1615. doi: 10.3964/j.issn.1000-0593(2009)06-1611-05MA Benxue, YING Yibin, RAO Xiuqin, et al. Research progress of hyperspectral imaging in nondestructive testing of fruit internal quality[J]. Spectroscopy and Spectral Analysis, 2009, 29(6): 1611-1615. doi: 10.3964/j.issn.1000-0593(2009)06-1611-05 陈全胜,赵杰文,蔡健荣,等. 利用高光谱图像技术评判茶叶的质量等级[J]. 光学学报,2008,28(4): 669-674. doi: 10.3321/j.issn:0253-2239.2008.04.011CHEN Quansheng, ZHAO Jiewen, CAI Jianrong, et al. Using high-resolution image technology to judge the quality level of tea[J]. Acta Optica Sinica, 2008, 28(4): 669-674. doi: 10.3321/j.issn:0253-2239.2008.04.011 邵瑰玮,付晶,陈怡,等. 基于图谱特征的复合绝缘子老化神经网络评估方法[J]. 高电压技术,2014,40(3): 861-867.SHAO Guiwei, FU Jin, CHEN Yi, et al. Neural network evaluation method of composite insulator aging based on map features[J]. High Voltage Engineering, 2014, 40(3): 861-867. 蓝磊,王汉良,文习山,等. 高温硫化硅橡胶加速电晕老化试验分析及其寿命预测[J]. 高电压技术,2012,38(4): 782-789.LAN Lei, WANG Hanliang, WEN Xishan, et al. Accelerated corona aging test analysis and life prediction of high temperature vulcanized silicone rubber[J]. High Voltage Engineering, 2012, 38(4): 782-789. 高岩峰,王家福,梁曦东,等. 交直流电晕对高温硫化硅橡胶性能的影响[J]. 中国电机工程学报,2016,36(1): 274-284.GAO Yanfeng, WANG Jiafu, LIANG Xidong, et al. Effect of AC/DC corona on properties of high temperature vulcanized silicone rubber[J]. Proceedings of the CSEE, 2016, 36(1): 274-284. CIGRE WG D 1.14. Material properties for non-ceramic outdoor insulation: state of the art, No. 255[R]. [S.l.]: CIGRE, 2004. 陈维荣,关佩,邹月娴. 基于SVM的交通事件检测技术[J]. 西南交通大学学报,2011,46(1): 63-67. doi: 10.3969/j.issn.0258-2724.2011.01.010CHEN Weirong, GUAN Pei, ZHOU Yuexian. Automatic incident detection technology based on SVM[J]. Journal of Southwest Jiaotong University, 2011, 46(1): 63-67. doi: 10.3969/j.issn.0258-2724.2011.01.010 孙永奎,余志斌,金炜东. 基于多重分形与SVM的高速列车运行状态识别方法[J]. 西南交通大学学报,2015,50(1): 7-12. doi: 10.3969/j.issn.0258-2724.2015.01.002SUN Yongkui, YU Zhibin, JIN Weidong. Recognizing running state of high-speed trains based on multifractal theory and SVM[J]. Journal of Southwest Jiaotong University, 2015, 50(1): 7-12. doi: 10.3969/j.issn.0258-2724.2015.01.002 李恒超,谭蓓,杨刚,等. 基于高光谱成像技术的绝缘子污秽度预测[J]. 西南交通大学学报,2019,54(4): 686-693.LI Hengchao, TAN Bei, YANG Gang, et al. Contamination degree prediction of insulators based on hyperspectral imaging technology[J]. Journal of Southwest Jiaotong University, 2019, 54(4): 686-693. 肖白,刘亚伟,施永刚,等. 基于主成分分析的中压配电网供电可靠性评估[J]. 电力自动化设备,2018,38(10): 7-12.XIAO Bai, LIU Yawei, SHI Yonggang, et al. Reliability evaluation of medium voltage distribution network power supply based on principal component analysis[J]. Electric Power Automation Equipment, 2018, 38(10): 7-12. 张春霞,郑钊,曹震,等. 基于改进的主成分分析法的地下水质量评价研究[J]. 绿色科技,2018(18): 50-53.ZHANG Chunxia, ZHEN Zhao, CAO Zhen, et al. Groundwater quality evaluation based on improved principal component analysis[J]. Journal of Green Science and Technology, 2018(18): 50-53. 李宝,王孟和,汪光胜,等. 基于高光谱的鲜桃叶片叶绿素含量检测[J]. 测绘工程,2018,27(10): 1-6.LI Bao, WANG Menghe, WANG Guangsheng, et al. Detection of chlorophyll content of peach leaves based on hyperspectral technology[J]. Engineering of Surveying and Mapping, 2018, 27(10): 1-6.