灰色关联度分析在变压器油色谱峰辨识中的应用  

Application of Grey Correlation Analysis in Chromatograph Peak Identification of Transformer Oil

作　　者：曹建[1] 范竞敏[1] 安晨光[2] 

机构地区：[1]中南大学物理科学与技术学院,湖南省长沙市410083 [2]北京华电云通电力技术有限公司,北京市丰台区100071

出　　处：《电网技术》2010年第7期206-210,共5页Power System Technology

基　　金：国家自然科学基金资助项目(50277039)~~

摘　　要：为弥补传统色谱峰辨识算法存在的缺陷,将灰色关联度技术应用到变压器油色谱峰辨识领域。以色谱学塔板理论为基础,先根据大量的色谱实验数据拟合一个匹配的高斯数据序列,然后将高斯数据序列沿着色谱数据滑动,滑动过程中求高斯窗口与相应色谱数据的灰色B型关联度,关联度大于设定阈值的数据段存在色谱峰时,该数据段色谱峰的位置对应滑到该处的高斯窗口的峰位。试验结果表明:该算法能非常准确地辨识色谱峰,对噪声、色谱峰宽变化不敏感,且与色谱峰的保留时间无关,具有非常好的抗干扰性和自适应性。To remedy the defects in conventional chromatographic peak identification algorithms, the grey correlation analysis is applied in chromatographic peak identification of transmission oil. According to the plate theory in gas chromatography, firstly based on plenty of chromatographic experimental data a Gaussian data sequence, which matches with the experimental data, is fitted; then the Gaussian data sequence is slide along the chromatographic data and during the sliding the grey B-type correlation degree between Gaussian window and corresponding chromatographic data is calculated, in the data segment where the correlation coefficient is greater than the predetermined threshold value the chromatographic peak exists and the position of the chromatographic peak in this data segment corresponds to the peak position of Gaussian window that moves to the place. Experimental results show that the proposed algorithm can identify chromatographic peaks accurately and is insensitive to both noise and change of the width of chromatographic peak, in addition, it is also independent of retention time of chromatographic peak, so it possesses excellent noise immunity and adaptability.

关 键 词：变压器油 色谱峰辨识 灰色B型关联度 

分 类 号：TM855[电气工程—高电压与绝缘技术]