基于拉曼光谱的变压器油中特征气体检测与数据分析研究

发布时间：2018-06-02 21:13

本文选题：电力变压器 + 拉曼光谱　；参考：《重庆理工大学》2017年硕士论文

【摘要】：电力变压器在电力系统中的地位举足轻重,为准确获取电力变压器的运行状态,及时对其进行有效的维护,本文选用了基于DGA(Dissolved Gas Analysis)的拉曼光谱检测方法,以变压器油中溶解故障特征气体为检测研究对象,通过拉曼仿真获取故障特征气体的拉曼特性,通过拉曼光谱线型分析确定更加符合实际情况的光谱数据处理方式,通过新拉曼特征参量的挖掘与应用提高在现有实验条件下拉曼定量分析的精度,为电力变压器运行状态的可靠评估打下坚实基础。首先,针对七种变压器故障气体的拉曼特征频谱选择,建立了基于密度泛函理论(DFT)中B3LYP方法的气体分子构型优化和拉曼频率特征仿真分析,结合密度泛函理论与实际检测确定七种故障气体的特征拉曼频率,为故障气体种类的快速判别及定量分析打下了基础。其次,以CH_4、C_2H_2气体为例,在对光谱数据进行背景基线扣除、平滑去噪处理等合理的数据处理后,应用GaussAmp线型拟合对其拉曼谱线进行拟合,并与Voigt线型拟合的结果进行了分析对比,发现在本文的实验条件下,对0.625ml/cm3浓度的CH_4气体,两种拟合方式的校正决定系数虽然接近,但计算的谱峰面积差值竟然接近250a.u·cm-1;同时用AIC信息准则和校正决定系数两种评判标准都判断出:相比目前应用广泛的Voigt线型拟合,GaussAmp线型在对变压器油中溶解气体CH_4、C_2H_2的拉曼谱线拟合中具有更佳的效果。而后,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,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GaussAmp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0.99251和0.98294,大于基于GaussAmp谱峰参数拟合的0.97952和0.95366。表明对于乙炔气体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。最后,以CH_4气体为例,应用非参数统计模型,将能反映拉曼散射强度的新发现的拉曼特征参量——峰度与原有拉曼特征参量——谱峰面积有效结合来进行拉曼定量分析,得到了良好的效果,能为电力变压器的运行状态的有效评估提供可靠的技术支持。
[Abstract]:The power transformer plays an important role in the power system. In order to obtain the operation state of the power transformer accurately and effectively maintain it in time, this paper selects the Raman spectrum detection method based on the DGA (Dissolved Gas Analysis), and takes the dissolved fault characteristic gas in the transformer oil as the detection research object, and obtains through the Raman simulation. The Raman characteristics of the fault characteristic gas can be used to determine the spectral data processing mode which is more in line with the actual situation through the analysis of the Raman spectrum line type. Through the mining and application of the new Raman characteristic parameters, the accuracy of the Raman quantitative analysis under the existing experimental conditions is improved, which lays a solid foundation for the reliable evaluation of the operation state of the power transformer. To select the Raman characteristic spectrum of seven kinds of transformer fault gases, the gas molecular configuration optimization and the Raman frequency characteristic simulation analysis based on the B3LYP method in density functional theory (DFT) are established, and the characteristic Raman frequency of the seven kinds of fault gases is determined with the density functional theory and the actual detection, which is a quick discrimination and determination for the type of the fault gas. Secondly, taking CH_4, C_2H_2 gas as an example, after deducting the background baseline of the spectral data, smoothing the noise processing and other reasonable data processing, the GaussAmp line fitting is used to fit the Raman spectra with the GaussAmp line fitting, and the results of the fitting of the Voigt line are analyzed and compared. It is found that under the experimental conditions of this paper, 0.6 25ml/cm3 concentration of CH_4 gas, the correction coefficient of the two fitting methods is close, but the calculated spectral peak area difference is close to 250a.u cm-1. At the same time, the AIC information criterion and the correction decision coefficient two criteria are all judged that the GaussAmp line is dissolved in the transformer oil compared to the widely used Voigt line fitting. The Raman spectrum fitting of gas CH_4 and C_2H_2 has a better effect. Then, in order to improve the accuracy of the quantitative analysis of dissolved gases in transformer oil based on the existing experimental conditions, from the descriptive statistics point of view, the new characteristic parameters in the Raman spectrum are excavated and the descriptive statistical parameters are found. According to the experimental data of Raman spectra of acetylene gas, the quantitative analysis method of spectral peak parameters based on GaussAmp fitting and the quantitative analysis method based on kurtosis are processed respectively, and the results are evaluated based on statistics, and the correlation coefficient based on kurtosis fitting is obtained. The positive determinant coefficients are 0.99251 and 0.98294 respectively, which are greater than the 0.97952 and 0.95366. based on the GaussAmp peak parameter fitting. It shows that the quantitative analysis of the kurtosis based on the acetylene gas is better than the current quantitative analysis of the Raman gas detection in the condition that the experimental data is not pretreated. Finally, the CH_4 gas is used as an example. A new Raman characteristic parameter, the Raman characteristic parameter, which can reflect the Raman scattering intensity, is effectively combined with the original Raman characteristic parameter, the spectral peak area, which can provide reliable technical support for the effective evaluation of the operation state of the power transformer.
【学位授予单位】：重庆理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM41

【参考文献】