期刊文献+

基于图像特征识别的绝缘纸老化状态评估 被引量:6

Aging State Evaluation of Insulating Paper Based on Image Feature Recognition
下载PDF
导出
摘要 针对现有绝缘纸状态检测方法的不足,提出一种新型无损检测方法,利用拍摄所得绝缘纸图像的纹理特征来评估其老化状态。首先,将不同老化阶段的绝缘纸进行图像采集和预处理,通过计算图像的灰度行程矩阵从而得到其纹理特征值;然后,利用相关分析筛选与聚合度相关性较高的特征,并在此基础上,利用支持向量机对所筛选特征表征绝缘纸老化状态的有效性进行验证;最后,采用多元回归分析法得到关键纹理特征与聚合度的拟合关系式,并采用实测结果进行验证。结果表明,绝缘纸的纹理特征与聚合度具有较好的拟合关系,预测值与实际值的误差率小于10%,验证了所提方法用于绝缘纸老化状态评估的可行性和有效性。该文方法配合伸缩式内窥镜可将理论方法推广到实际应用,为未来变压器检修带来便捷性。 To deal with the deficiencies of existing insulating paper state detection methods,a new non-destructive detection method is proposed,in which the texture characteristics of the captured insulating paper images are adopted to evaluate its aging state.Firstly,the insulating papers in different aging stages are collected and preprocessed,and its texture feature value is obtained by calculating the gray-level run-length matrix of the image.Then,correlation analysis is used for filtering the features that are highly correlated with the degree of polymerization.Based on this,the support vector machine is used for verifying the effectiveness of the selected features to characterize the aging state of the insulating paper.Finally,the multiple linear regression method is used to obtain the fitting relationship between the texture features and the degree of polymerization,and it is verified by the actual measured results.The results show that the measured texture characteristics of insulating paper have a good fitting relationship with the degree of polymerization,and the error rate between predicted value and actual value is less than 10%,which verifies the feasibility and effectiveness of this method for evaluating the aging state of insulating paper.The method,combined with the telescopic endoscope,can extend the theoretical method to practical application,and it can bring some convenience for the future transformer maintenance.
作者 崔家齐 董海鹰 李帅兵 康永强 CUI Jiaqi;DONG Haiying;LI Shuaibing;KANG Yongqiang(School of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;School of New Energy and Power Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China)
出处 《高电压技术》 EI CAS CSCD 北大核心 2022年第2期636-643,共8页 High Voltage Engineering
基金 国家自然科学基金(52067014) 兰州交通大学青年基金(2019029)。
关键词 无损检测 图像纹理特征 灰度行程矩阵 支持向量机 多元回归分析 non-destructive testing image texture characteristic gray-level run-length matrix support vector machine multiple linear regression analysis
  • 引文网络
  • 相关文献

参考文献14

二级参考文献207

共引文献493

同被引文献199

引证文献6

二级引证文献69

;
使用帮助 返回顶部