Research On Aging Diagnosis Of Oil-paper Insulation Based On The Laser Raman Spectroscopy

Aging of oil-paper insulation in power transformers will cause equipment failure and affect the reliable operation of power system.As a result,accurate diagnosis of the aging degree of oil-paper insulation materials and timely grasp of the insulation aging state of the power equipment can ensure the safe operation of the equipment and the power grid.During the process of electric or thermal aging,oil-paper insulation materials will produce various substances related to the aging state,such as furfual,methanol,acetone,CO and CO₂,and dissolved in the oil.At present,the aging diagnosis of running transformers is mainly based on the detection and analysis of these aging characterstics,and the corresponding threshold value is used for laboratory judgement.There is a problem that a single feature quantity needs different equipment analysis and cannot be used for effective on-site diagnosis.Laser Raman technology has some advantages in the field of material composition analysis and state diagnosis.This paper relies on the guideline project of the State Grid Corporation of China?SGS00200FCJS1700228?to carry out the research on diagnosis of oil-paper insulation aging based on the laser Raman spectroscopy.This paper studies the influence of electrical and thermal stress on the oil-paper insulation,determines the distribution and attribution of Raman characteristic peaks of aging features,analyzes the relationship between spectral features and aging degree,and introduces a machine learning classifier to establish an aging diagnosis model,and verifies the discriminant ability of the model to the aging state of the actual transformer.The research contents and the main results obtained are as follows:?1?Considering the combined effect of electric field and temperature,the accelerated aging test of oil-paper insulation was carried out in laboratory,and oil-paper insulation samples with different aging types and aging stages were prepared.The change rules of chemical characteristics in different aging types during aging process were studied,and the relationship between the degree of polymerization?DP?of insulating paper,furfural content in oil,moisture content in oil and dissolved gas in oil and aging degree were analyzed.At the same time,the effect of thermal stress and electrical-thermal stress on the aging of oil-paper insulation was explained from a microscopic perspective.?2?The characteristic peak distribution and vibration attribution of main dissolved characteristics?furfural,methanol,acetone,CO and CO₂?during the aging process of oil-paper insulation were determined.Based on the principle of Raman spectroscopy,the Raman spectrum detection platform of insulating oil was built in laboratory,and the raw data of Raman spectrum at different aging stages were obtained.The three point sliding window average method,the five point three smoothing method and polynomial fitting method were used for data preprocessing respectively.?3?The aging samples were divided into four stages according to DP.The kernel principal component analysis was used to reduce the dimension of the spectral data.The accumulative contribution rate of the first 8 principal components was 95.81%,the spectral data fell to 8 dimensions,the spectral characteristics reflecting the aging state were extracted,and the intrinsic relationship between the new feature quantity and the aging degree was analyzed.The partial least square method was used to screen out the singular samples in the data samples,and the training samples of the diagnostic model were selected by Kennard-Stone.?4?Based on the extracted principal component features,three machine learning classifiers?kernel Fisher discriminant analysis,kernel function support vector machine,radial basis kernel function neural network?were introduced,and parameters were optimized by the optimization algorithm.Three aging diagnosis models were built to identify different aging stages.The diagnostic accuracy of KSVM is the highest after grid search.Finally,this paper compares the results of Raman spectrum diagnosis and traditional characteristic diagnosis of the 50 typical transformer oil samples,and verifies the discriminating ability and generalization ability of the oil-paper insulation aging diagnosis method based on Raman signal.The research shows that the aging diagnosis method of oil-paper insulation based on Raman spectroscopy has good aging discrimination ability,which lays the foundation for the fast diagnosis of the aging condition of transformer oil-paper insulation.