Status Assessment Of Aging Condition Of Oil-Paper Insulation Based On Time-Domain Dielectric Spectroscopy And Texture Analysis

The oil-paper insulation system in the power transformer is affected by a combination of various factors such as temperature,moisture,acid,and electromagnetic in the course of operation for a long time,and its insulation performance will continue to deteriorate,thereby affecting the service life of the transformer.Therefore,how to extract effective and accurate feature quantities has important theoretical significance and practical engineering value for monitoring the operating status of oil-immersed power transformers and prolonging their use time.Combining laboratory conditions this paper builds oil-paper composite insulation accelerating thermal aging test platform in the laboratory.Polarization/depolarization current tests were performed on the obtained samples,and the time domain dielectric properties of oil-paper composite insulation in different aging states were analyzed.The analysis of the trend of the PDC curve of oil-paper composite insulation shows that: The thermal aging of the oil-paper composite insulation degrades the cellulose structure in the insulating paper.The degradation product further accelerates the aging of the insulating paper,accelerates the degradation of the cellulose,breaks the molecular chains in the cellulose,and generates more internals in the insulating paper.In the gap,the insulating oil can easily enter the interior of the insulating paper,thereby forming more oil-paper interfaces,resulting in enhanced interfacial polarization,resulting in the overall upward movement of the PDC curve.The analysis of the second-differential time-domain dielectric spectrum of the depolarization current of the oil-paper insulation sample reveals that as the thermal aging time increases,the second-differential time-domain dielectric spectrum curve shifts upward overall.Peaks appear at the later stage of thermal aging,which are very sensitive to aging time.Combined with Meyer wavelet analysis algorithm,the original secondary time-domain dielectric spectrum is improved,and three main polarization processes of oil-paper insulation can be directly obtained from the improved time domain dielectric spectrum: oil,paper,oil-paper interface.The polarization relaxation in oil is very short,generally completed in the initial stage of depolarization current,and the time constant is within 10 s.The polarization time constant of the oil-paper interface is about 700 s,and peaks with larger time constant represent the polarization process inside the insulating paper.The relationship between the different peaks of the improved time-domain spectroscpy and the aging time of each part of the oil paper is fitted by a formula,and the fitting relationship is good.The spectrum peak can be considered as the characteristic parameters of different parts of oil-paper insulation.Image processing is applied to the evaluation of the ageing degree of oil-paper composite insulation.After preprocessing of the oil-paper composite insulation image,a gray-level co-occurrence matrix is obtained and its characteristic parameters are calculated.According to the correlation of feature parameters,eight effective feature quantities are extracted,and two new feature quantities are extracted using principal component analysis.Using different algorithms(supervised and unsupervised algorithms)to verify the reliability of the extracted features,the results show that: the classification accuracy using ELM or ANN is quite high,and the different classification algorithms have a slight effect on the results;the use of k-means clustering algorithm in the classification of aged insulation paper samples also has good results,and the best results were obtained when using 6 clusters.Using multivariate regression analysis to predict the aging state of insulating paper,the results show that if the selected feature quantity is appropriate,multivariate regression analysis can well predict the aging state of insulating paper.