Study On The Detection Of Polymerization Degree Of Trnasformer Insulating Paper Based On Diffuse Reflection Spectroscopy

The polymerization degree of insulation paper can effectively reflect the aging degree of power-transformer, but the viscosity method is complex and time-consuming to detect the polymerization degree of insulation paper. To solve this problem, this paper provides a rapid method to detect the polymerization degree of insulation paper based on spectroscopy.In this paper 54 of transformer insulation paper with different degrees of polymerization were aged with oxygen and another 54 of transformer insulation paper with different degrees of polymerization were aged in transformer oil without oxygen. It is found that the aging mechanism of the twokind of aging way were different. The process of aging way with oxygen was faster than the one without oxygen.These aged insulation paper were scanned by diffuse reflectance spectroscopy. The spectral range is 198-2547 nm. To establish the regression model of degree of polymerization and spectral information by using partial least squares.The experimental results show that:(1)Scanning distance, average frequency of spectrum, pretreatment method, modeling method and wavelength range were studied. The best parameters were ch osed. Scanning distance: 0mm, average frequency of spectrum: 5, pretreatment meth-od: 8 point smoothing, modeling method: linear partial least squares regression, wavelength range: 350-1000 nm, 1100-1600 nm, 1800-2200 nm.(2)There is a good linear relationship between the degree of polymerization and spectral information of oxygen-aged insulation paper. The performance of the model is good, correlation coefficient of calibration of RC is 0.9742, the root mean square error of prediction RMSEP is 33.5. The detection precision and accuracy can meet the basic requirements for engineering applications.(3)The performance of non-oxygen-aged insulation paper(with transformer oil) model is poor, the calibrated correlation coefficient RC is 0.8629, the root mean square error of prediction RMSEP is 127. Transformer oil can cause interference,the more insulating oil in insulation paper,the more interference exist.After removing the interference of transformer insulating oil, the performance of the model is greatly improved, the correlation coefficient become 0.9745, the root mean square error of prediction RMSEP become 33.9.(4)Cluster analysis was performed on three kinds of insulation paper, oxygen-aged insulation paper and non-oxygen-aged insulation paper(without transformer oil) are in the same class, and non-oxygen-aged insulation paper(with transformer oil) is in another. The aging mechanism of oxygen-aged way and non-oxygen-aged way were different, but the regression relationship between the degree of polymerization and spectral information is basically the same.In summary, the spectral method can detect the degree of polymerization for oxygen-aged insulation paper accurately. After removing the transformer insulating oil, the spectral method can also detect the degree of polymerization for non-oxygen-aged insulation paper accurately.