Evaluation Study Of Degradation Of Oil-immersed Insulation In Transformer Hot Spot Region Based On Frequency Dielectric Response Method

The overall lifespan of the insulation system in transformers is determined by the aging status of the oil-paper insulation.The distribution of the aging status of the oil-paper insulation is non-uniform due to the temperature gradient created by the arrangement of windings and the heat dissipation of the iron core during operation.The hot spot area of the winding exhibits the most severe oil-paper insulation aging and poses the highest operating risk,thus being regarded as the "weak area" of the entire oil-paper insulation system.Currently,the evaluation of oil-paper insulation aging predominantly depends on indirect methods such as chemical and electrical parameters.However,these traditional measures are only capable of characterizing the "average aging status" of the oil-paper insulation,rendering them insufficient to reflect the aging status of the hot spot area.In recent years,the dielectric response method has garnered attention for its ability to carry rich insulation information and conduct non-destructive measurements.However,research on the characterization and evaluation of oilimmersed paper insulation aging status using the dielectric response method primarily focuses on uniform aging forms and lacks understanding of hot spot aging forms.Therefore,this paper aims to introduce a new method utilizing the dielectric response technique to quantitatively characterize the insulation status of the hot spot area of transformer oil-immersed insulation.The main innovations of this paper include:1.A model for obtaining insulation information in the hot spot area of transformers is proposed.By combining Debye relaxation theory and XY model theory,a "modified XY model" considering the non-uniform aging effect of the insulation system is established to construct the quantitative relationship between the overall FDS of the insulation system and the insulation information in the hot spot area under non-uniform aging conditions.Based on this,a computation model is proposed using the "multi-constraint non-dominated sorting genetic algorithm(MC-NSGA)",which enables the acquisition of information on the aging and moisture state of the insulation in the hot spot area of the transformer.2.A method for obtaining characterization parameters of paper insulation status in transformer hot spot areas is proposed.Based on decoupling analysis of FDS data,the logarithmic-derivative spectroscopy(LDS)is reported for analyzing the FDS data of the transformer’s hot spot area and extracting insulation status feature parameters.This method can effectively eliminate the covering effect of conductivity and electrode polarization on low-frequency relaxation polarization information,explore the characteristics of low-frequency relaxation polarization behavior of oil-paper insulation,and achieve accurate analysis of insulation aging and moisture status.3.A simulation database construction model considering non-uniform aging effects for FDS is proposed.The construction of the insulation feature parameter database is an important prerequisite for intelligent assessment of insulation status.However,there is currently a lack of a method for constructing a database for evaluating the oil-immersed insulation status of the hot spot area of transformers.Therefore,a simulation model based on swarm intelligence optimization is proposed to simulate the FDS data of composite oil-paper insulation systems under non-uniform aging conditions,providing a data basis for the development of subsequent status assessment models.4.An intelligent assessment model for insulation aging and moisture status in the hot spot area of transformers is proposed.By using the FDS simulation database to construct a feature parameter data set that characterizes the insulation status in the hot spot area,an intelligent classifier for assessing the insulation status in the hot spot area is constructed by combining convolutional neural networks(CNN)and optimized KNN algorithms.Furthermore,the accuracy and reliability of the model for analyzing the aging and moisture status of insulation in the hot spot area of transformers are verified by laboratory samples and field transformer status assessment experiments.