Quantitative Evaluation Research On Moisture Content Of Oil-paper And Aging Degree Of Transformer Insulation

Under the stress of heat,mechanical oscillation,electric field and other factors the performance of transformer oil-paper insulation system gradually deteriorates.The performance of aged insulating oil can be improved by oil filtering or changing operation,while replacement for insulation paper and cardboard is more difficult,therefore,the aging degree of the transformer solid insulation is the key factor to determine the remaining life of transformer.Moisture is a kind of aging product which in return accelerates the aging of insulating oil and paper.How to effectively distinguish the influence of aging and moisture on frequency domain dielectric spectroscopy,and for the case of aging and moisture coexistence,how to take an effective way to separately assess the degree of aging and moisture content of oil-paper insulation,is the key to the frequency domain dielectric response method better applied to the evaluation of field transformer insulation state.In this paper,the oil-immersed insulating paper samples with different aging degree and moisture content were prepared by accelerated thermal aging test and moisture absorption test in laboratory,and tested the dielectric loss factor spectrum of the samples with different insulating state.The test result showed that in the case of low moisture content in paper the influence of moisture and aging on frequency domain dielectric spectroscopy could be distinguished by the difference of impacted frequency band.In the characteristic frequency band affected by moisture,the integral value of the dielectric loss factor to frequency was taken as a characteristic parameter to characterize the moisture content of oil-immersed insulating paper,and the quantitative relationship between the characteristic parameter and the moisture content of the test samples was established.Then,extracted a characteristic parameter which could effectively characterize the aging degree of oil-immersed insulating paper in the case of low moisture content in paper,and established the quantitative relationship between the characteristic parameter and the aging degree of the test sample in low moisture content.Based on the proposed "moisture compensation factor" quantitatively analyzed how moisture content influences the aging assessment of oil-paper insulation at high moisture content in paper.Finally,considering the influence of moisture content deduced an evaluation method for oil-immersed insulating paper aging degree by using the fitting relationship of "water compensation factor" between DP and Km.c.To solve the problem that the influence of aging and moisture on frequency-domain dielectric spectroscopy is difficult to be distinguished and the mapping relationship between frequency-domain dielectric spectroscopy and aging degree of oil-paper insulation and moisture content is complex,which leads to the characteristic parameters characterizing moisture and aging are difficult to be extracted.The number of oil-immersed insulating paper samples with different insulation state was expanded in the laboratory,and tested its frequency-domain spectrum and impedance phase frequency spectrum.Then,the multi-output support vector regression algorithm was used to excavate the complex mapping relationship between moisture content,aging degree and its frequency domain dielectric spectroscopy,and then used its generalization ability to predict the moisture content and aging degree of oil-paper insulation.Combined with the cluster analysis results based on self-organizing neural network,the result showed that multi-output support vector regression algorithm can effectively predict the moisture content and the aging degree of the oil-immersed insulating paper,when moisture content of insulating paper is less than 3%.This paper introduced the wiring method for field transformer frequency domain dielectric spectroscopy test,and obtained the FDS frequency spectrum of the insulating paperboard from the measured FDS frequency spectrum of the main insulation structure of field transformer by using X-Y model.To eliminate the influence of the test temperature on frequency domain dielectric spectroscopy,the FDS spectrum curve at the test temperature was shifted to the reference temperature by means of the "temperature shift" principle.Then,evaluation process for the methods of dielectric loss integral characteristic parameter and multi-output support vector regression algorithm were established to evaluate the solid insulation state of field transformer.Finally,verified the effectiveness of the two methods to evaluate the solid insulation state of field transformer,through frequency domain dielectric spectroscopy testing results of two transformers off-line for maintenance.