Aging Stage Diagnose Of Oil-paper Insulation Based On Raman Spectral Similarity

Oil-paper insulation is the main insulation of large power transformers.It is of great significance to determine the aging degree of oil-paper insulation accurately and quickly to ensure the safe and stable operation of equipment and power grid.However,at present,the aging diagnosis of large transformer oil-paper insulation is mainly based on the detection and analysis of the characteristic amount of the aging products such as furfural dissolved in the insulation oil,and the corresponding reference data provided by the electric power industry guidelines are analyzed and determined.There are some problems with this method.Specific analysis tools and means are needed to analyze a single characteristic quantity,while multiple characteristic quantities require multiple devices.Therefore,the transformer can be effectively diagnosed in the field.Using Raman detection method to test the transformer oil-paper insulation system does not need a variety of equipment,can be directly tested in the field,has certain advantages.In this paper,based on Raman detection technology,the aging characteristics in oil paper insulation system analysis,design of the transformer oil paper insulation system thermal accelerated aging experiment,analysis and judgment of transformer insulation aging of specific product types and content in the oil,on the Raman spectroscopy to determine the aging characteristic peak distribution features of the Raman and belonging.A point-by-point matching similarity of the full spectrum Raman spectrum of the transformer has been performed subsequently,and the experiment proves that the method of similarity analysis used in this paper has a good ability to judge the aging stage of the transformer.The main research work and achievements are as follows:1)The most important factor for transformer aging is temperature.When the equipment runs in a high temperature environment for a long time,it will accelerate the aging of oilpaper insulation system.In this paper,the accelerated thermal aging experiment of transformer oil-paper insulation system is set up in the laboratory.Transformer oil in different aging stages will produce specific aging products,and the content and types of each aging product have specific rules.The relationship between the polymerization degree of insulating paper,furfural content in oil,water content in oil and dissolved gas content in oil and aging degree was experimentally studied.The effect of accelerated thermal aging on the aging of oiled paper insulation was explained from the molecular structure of the aging products.2)The analysis of the insulating oil using Raman spectroscopy analysis platform for the lab building,through to look at the different stages of aging of transformer oil sample analysis,using the simulation software of Gaussian 09 w simulation several characteristic peak aging characteristics of products,including furfural,CO,CO₂ and methanol,from theory and actual measurement of the Raman characteristics of the aging characteristics of the product and the molecular vibrational mode,and has carried on the contrast.The original Raman spectral data obtained from the platform at different aging stages were preprocessed by three point sliding window averaging method,five point cubic smoothing method and polynomial fitting method,which made the subsequent analysis easier.3)The most direct measure to distinguish the aging stage of transformer oil paper insulation system is the degree of polymerization of insulating paper.According to the degree of polymerization of insulating paper,it can be divided into four stages: unaging,early aging,middle aging and late aging.Due to certain errors and abnormal samples in sample detection of Raman detection platform,partial least square method was used to eliminate abnormal samples generated by interference in the experiment,and KS?Kolmogorov-Smirnov?algorithm was used to select the training samples to build the diagnostic model.4)Raman spectroscopy to detect transformer oil respectively,and the degree of polymerization of transformer insulation paper,combining both data to find its corresponding relationship,establish a general aging phase identification of library,and uses the appropriate image similarity matching algorithm for different spectral image matching using the Euclidean distance,chi-square distance and jenson-shannon divergence on the transformer insulation Raman spectrum matching.Finally,100 sets of transformer oil samples in actual operation are selected for actual detection,and the discriminant ability of the method for diagnosing the aging state of oil-paper insulation based on Raman signal is verified according to the judgment results.The results show that the method based on Raman spectrum can accurately and quickly identify the aging status of transformer oil-paper insulation.