基于峰度的变压器油中特征气体拉曼光谱分析

发布时间：2019-02-28 11:58

【摘要】：为了在现有实验条件的基础上提高拉曼光谱检测用于变压器油中溶解气体定量分析的精度,进而更加准确地获取变压器的工作状态,从描述性统计学角度入手,挖掘拉曼光谱中新的特征参量,发现描述性统计参量峰度,可用作拉曼光谱分析中的新参量。针对乙炔气体的拉曼光谱实验数据,分别用基于Gauss Amp拟合的谱峰参数定量分析方法和基于峰度的定量分析方法进行了处理,并对结果进行了基于统计学的评价。得到基于峰度拟合的相关系数与校正决定系数分别为0.992 51和0.982 94,大于基于Gauss Amp谱峰参数拟合的0.979 52和0.953 66。表明基于峰度的定量分析在未对实验数据预处理的情况下得到了比目前拉曼气体检测定量分析更好的效果。
[Abstract]:In order to improve the accuracy of Raman spectrum detection for quantitative analysis of dissolved gases in transformer oil on the basis of existing experimental conditions, and to obtain the working state of transformer more accurately, this paper starts with descriptive statistics. The new characteristic parameters in Raman spectrum are excavated and the descriptive statistical parameter kurtosis is found which can be used as a new parameter in Raman spectroscopy. According to the experimental data of Raman spectra of acetylene gas, the quantitative analysis method based on Gauss Amp fitting and the quantitative analysis method based on kurtosis were used to analyze the spectral peak parameters, and the results were evaluated based on statistics. The correlation coefficient and correction determination coefficient based on kurtosis fitting are 0.992 51 and 0.982 94, respectively, which are higher than 0.979 52 and 0.953 66 based on Gauss Amp peak parameter fitting. It is shown that the quantitative analysis based on kurtosis is better than the quantitative analysis of Raman gas detection in the absence of pretreatment of experimental data.
【作者单位】：重庆理工大学电气与电子工程学院;重庆大学输配电装备及系统安全与新技术国家重点实验室;重庆市能源互联网工程技术研究中心;
【基金】：国家自然科学基金青年科学基金(51107155) 重庆市教委科学技术研究项目(KJ1400917) 重庆市科委研究项目(CSTC2013jcyj A900019)~~
【分类号】：TM41

【参考文献】