基于光谱指数特征的玻璃绝缘子表面污秽层湿润度定量反演

Quantitative Inversion Moisture Content in Pollution Layer on the Surface of Glass Insulator Via Spectral Indexes

准确开展污秽层湿润度检测对于评估污闪风险、研究污闪机理具有重要意义。本文研究了不同湿润度下玻璃绝缘子污秽样品反射光谱特性,分析了水分对反射特性的影响机制,构造了表征湿润度的光谱指数,建立了基于光谱指数的湿润度检测方法。结果表明,水分的吸收、散射以及污秽层状态的改变是引起湿润污秽层反射率下降和反射光谱形状变异的主要原因;由816.88和591.80 nm组成的波段比值指数与湿润度最大相关系数绝对值达到了0.986 1;由差值-归一化-波段比值指数构成的稳健回归模型湿润度预测均方根误差仅为0.051 29,拟合优度达到了0.98;光谱图像数据的“图谱合一”特征实现了污秽层湿润度的可视化呈现。本研究为污闪过程机理研究和绝缘子设计优化等研究提供了重要手段。

Conducting accurate moisture measurement of the pollution layer is of paramount significance for assessing pollution flashover risk and investigating the pollution flashover mechanism.In this paper,the reflectance spectral characteristics of glass insulator samples under various moisture conditions were examined,and the mechanism influencing the reflectance characteristics by water was analyzed.Three spectral indexes representing wetness were constructed,and a moisture detection method based on the spectral indexes was established.The results indicate that the absorption and scattering of water and the alteration of the state of the pollution layer are the primary causes for the reduction of the reflectance and the variation of the reflectance spectra.The absolute value of the maximum correlation coefficient between the band ratio index of 816.88 nm and 591.80 nm and the moisture content(MC)attains 0.9861.The root-mean-square error of MC prediction in the robust regression model composed of the difference normalization-band ratio index is merely 0.05129,and the goodness of fit reaches 0.98.The integration of"image-spectral features"of spectral image data enables the visualization of the MC.This research offers an essential means for the study of the mechanism of the pollution flashover process and the optimization of insulator design.