Research On Assessment Of Insulation Aging Status Of Transformer Oil-paper Based On RVM

As an indispensable part of the power grid,oil-filled power transformers play a crucial role in ensuring stable power supply.Being a crucial component of the transformer,oil-paper insulation is susceptible to aging and dampness in a long-term working environment,which seriously threatens the reliability of insulation.To solve the problem that the suspension potential in the test phase causes the test result to be too large,the thesis uses a double switch to force grounding.In this thesis,a double switch is used to force grounding,and the conversion of the RVM / PDC test circuit is realized by the control logic circuit.At the same time,accelerated thermal aging and dampness tests were carried out on oil-paper insulated samples under laboratory conditions.Multiple sets of oil-paper insulated aging and dampness samples were prepared,and RVM tests were conducted at different test temperatures.At the same time,observing the sample,the test temperature has little effect on the variation of the test characteristic of the aging and dampness oil-paper insulation model,and only the effect of the temperature correction factor on the curve should be considered.Regarding the RVM curves obtained from the tests,this study extracted feature parameters related to the oil-paper insulation state in RVM,such as peak voltage,initial slope,and center time constant,to construct a pattern recognition library based on the feature parameters of the oil-paper insulation through RVM.Since conventional insulation aging assessment methods cannot quantitatively evaluate the aging state of the tested model using a single type of test data,this study performed theoretical analysis on the recovery voltage feature parameters and proposed a classification method for characterizing the insulation aging state of oil-paper insulation based on an improved TOPSIS algorithm.The PSO-ELM neural network algorithm was used to adjust the output layer to achieve the prediction of the model aggregation degree and moisture content.However,the experiments showed that the accuracy of moisture content prediction was higher,while the aging degree result of the prediction had a larger deviation.Thus,this study further employed the SVM model,which considers the effect of moisture content and eliminates the influence of gradient explosion or disappearance,to accurately evaluate the paper insulation aging degree(aggregation degree).The analysis results demonstrated that the predicted aggregation degree of the oil-immersed paperboard was within 4% of the actual aggregation degree,and the insulation classification accuracy reached 96%.The predicted results met the needs of engineering testing.