基于主成分回归分析的SF_6分解组分红外光谱定量

Infrared Spectrum Quantitative Method of SF6 Decomposition Components Based on Principal Component Regression Analysis

SF6气体分解组分的定量分析对于气体绝缘组合电器(gas insulated switchgear,GIS)设备的在线监测与故障诊断具有重要的意义。基于朗伯比尔定律进行气体红外定量分析时存在吸收系数的难确定性和多组分检测的繁琐性问题,为解决此问题,采用了主成分回归分析的红外定量算法,并选取SO2F2、SOF2、SO23种分解组分及其特征吸收峰作为研究对象。该算法分别以光谱数据与气体体积分数作为输入和输出,建立了吸收光谱图与浓度信息的直接对应关系。试验配制了3种特征组分的含量样本,由训练样本获得光谱数据并依据算法建立了红外定量预测模型。通过比较主成分回归分析算法和拟合法的红外定量效果,表明了回归算法模型定量精度高于拟合法。检验样本和实测数据的预测结果都表明,该算法和预测模型具有很好的适用价值。

Infrared quantitative analysis of SF6 gas decomposition components is of great significance for on-line monitoring and fault diagnosis of gas insulated switchgear（GIS） equipment.Some problems,such as uncertainty of absorption coefficient and fussy calculation of multi-composition detection,existed in gas infrared quantitative analysis,when the Beer-Lambert law was applied.Consequently,an infrared quantitative algorithm based on principal component regression analysis was presented,moreover,three components SO2F2,SOF2,SO2 and their characteristic absorption peaks were selected as research objects.Spectra data and gas concentration were regarded as imports and exports of the algorithm,so the direct corresponding relation between absorption spectrogram and gas concentration was established.Concentration samples of three Characteristic components were prepared,afterwards,spectra data were obtained and the infrared quantitative forecast model was established on the basis of the algorithm.The results showed that the accuracy of the infrared quantitative model is higher than that of the fitting method,by comparing infrared quantitative effect.Forecast results of testing samples and real data indicated that the algorithm and the forecast model have a very good applicable value.