色谱

色谱 ›› 2014, Vol. 32 ›› Issue (9): 1019-1024.DOI: 10.3724/SP.J.1123.2014.05008

• 研究论文 • 上一篇    

一种变压器油色谱峰识别算法的设计

欧林军1, 曹建1,2   

  1. 1. 中南大学物理与电子学院, 湖南 长沙 410083;
    2. 北京华电云通电力技术有限公司, 北京 100071
  • 收稿日期:2014-05-12 修回日期:2014-07-09 出版日期:2014-09-08 发布日期:2014-08-30
  • 通讯作者: 曹建
  • 作者简介:曹建，caojian1955@126.com.

A peak recognition algorithm designed for chromatographic peaks of transformer oil

OU Linjun1, CAO Jian1,2   

  1. 1. School of Physics and Electronics, Central South University, Changsha 410083, China;
    2. Beijing Huadian Yuntong Power Technical Co., Ltd., Beijing 100071, China
  • Received:2014-05-12 Revised:2014-07-09 Online:2014-09-08 Published:2014-08-30

PDF

327

被引次数

2

摘要：

在变压器油色谱峰识别领域，传统的一阶导数法需要斜率阈值来实现色谱峰识别，因此自动化程度低及容易失真。针对这些缺点，本文在一阶导数法的基础上进行了改进，将迭代移动平均及归一化分析技术应用到色谱峰识别中，通过对信号曲线及方波曲线多次迭代移动平均确定最优的归一化识峰参数，结合色谱峰的绝对保留时间及识峰窗口实现对色谱峰的准确识别。实验结果表明：该算法可以准确识别色谱峰，对噪声、色谱峰的峰宽及峰形变化不敏感，具有很强的自适应性，满足变压器油中气体在线监测装置的现场使用要求。

关键词: 变压器油, 迭代移动平均法, 归一化识峰参数, 色谱峰识别

Abstract:

In the field of the chromatographic peak identification of the transformer oil, the traditional first-order derivative requires slope threshold to achieve peak identification. In terms of its shortcomings of low automation and easy distortion, the first-order derivative method was improved by applying the moving average iterative method and the normalized analysis techniques to identify the peaks. Accurate identification of the chromatographic peaks was realized through using multiple iterations of the moving average of signal curves and square wave curves to determine the optimal value of the normalized peak identification parameters, combined with the absolute peak retention times and peak window. The experimental results show that this algorithm can accurately identify the peaks and is not sensitive to the noise, the chromatographic peak width or the peak shape changes. It has strong adaptability to meet the on-site requirements of online monitoring devices of dissolved gases in transformer oil.

Key words: chromatographic peak identification, moving average iterative method, normalized peak identification parameters, transformer oil

中图分类号: