全文下载：2013040419
刘砚菊， 潘胜， 于洋， 陈亮
（沈阳理工大学 信息科学与工程学院, 辽宁 沈阳 110159）
摘要： 将盲信号分离算法与光声光谱检测平台应用到变压器绝缘油中混合气体的检测，能降低对激励光源精度的要求，减少成本，在相同光源的情况下可以提高检测精度。利用盲信号处理中的负熵最大法对CO2、C2H6、C2H23种红外特征吸收峰中心波长特别接近的主要故障气体进行了盲源分离。实验结果表明，与单纯的光声光谱法进行比较，利用盲信号分离算法能更好地检测出绝缘油中的混合气体的种类和含量，具有较高的检测精度。
关键词： 混合气体； 盲信号分离； 光声光谱； 负熵最大法
中图分类号: TN 919.5文献标志码: A
Method of Blind Signal Separation of Mixed Gases in Transformer Insulation Oil
LIU Yan-ju, PAN Sheng, YU Yang, CHEN Liang
(School of Information Science and Engineering, Shenyang Li gong University,Shenyang110159,China)
Abstract: In this paper, Blind Signal Separation (BSS) method is applied to detect the gases in transformer insulation oil. The photoacoustic spectroscopy testing platform and Blind Signal Separation algorithm are employed to reduce the precision requirements of the light source and save cost. At the same time, the detection precision can be improved under the same strength of light. The Maximum Negative Entropy Method of blind signal processing is used to separate the main fault gases in CO2, C2H6 and C2H2, whose center wavelength of infrared absorption peaks features is especially close. The experimental results shows that Blind Signal Separation algorithm can detect the varieties and contents of mixed gases compared with the simple photoacoustic spectroscopy method, and can improve the detection precision.
Key words: mixed gases; blind signal separation; photoacoustic spectroscopy; maximum negative entropy method
收稿日期：2012-11-12
作者简介：刘砚菊（1965—），女，教授.
