污闪是威胁电网安全可靠运行的重要原因之一, 污秽类型差异将直接影响闪络电压大小。因此, 及时掌握绝缘子污秽类型信息对预防污闪有重要作用。为此提出了一种基于SAM-ED光谱匹配的绝缘子污秽类型检测方法。采集不同污秽类型样本高光谱数据, 经黑白校正及多元散射校正(MSC)去除噪声等干扰因素;利用竞争自适应重加权采样法(CARS)对光谱数据进行特征选取, 分别在特征波段和全波段范围内通过SAM-ED光谱匹配法将测试组样本光谱与参考光谱进行匹配, 根据匹配结果对样本进行分类;实验结果表明: 相比于光谱角匹配法和最小距离法, SAM-ED光谱匹配法检测效果更好;基于全波长数据进行SAM-ED光谱匹配准确率可达95%, 基于特征波长数据进行SAM-ED光谱匹配准确率可达98.33%。

Fouling flash is one of the important reasons that threaten the safe and stable operation of power system, and the difference of fouling type will directly affect the size of flashover voltage. Therefore, timely information on insulator fouling type plays an important role in preventing fouling. To this end, a SAM-ED spectral matching based insulator fouling type detection method is proposed. The hyperspectral data of different fouling type samples are collected, and the noise and other interference factors are removed by black and white correction and multiple scattering correction (MSC); the spectral data are selected by competitive adaptive reweighted sampling (CARS), and the samples are matched with the reference spectra by SAM-ED spectral matching method in the characteristic band and full band range, respectively, and the samples are classified according to the matching results. The experimental results show that the SAM-ED spectral matching method is more effective than the spectral angle matching method and the minimum distance method, and the accuracy of SAM-ED spectral matching based on the full wavelength data can reach 95%, and the accuracy of SAM-ED spectral matching based on the characteristic wavelength data can reach 98.33%.<通讯作者>刘益岑(1986-), 男, 硕士, 高级工程师, 研究方向为电网防灾减灾及在线监测等