| As the hub equipment of transmission line and distribution network in power network,oil immersed transformer has an important influence on the stability of power network.The main insulation of oil immersed transformer is composed of oil-paper insulation.The insulation state of oil paper insulation also determines the life of transformer,and with the increase of transformer operation time,oil-paper insulation gradually aging under the action of electricity and heat,coming into being a large number of aging products,and the appearance of these impurities will aggravate partial discharge(PD),and the occurrence of PD will accelerate the aging of oil-paper.In view of the situation that the life of oil-paper insulation is decreased and the complex partial discharge is caused by different stages of thermal aging,an experimental platform for PD occurs in oil-paper insulation is built based on pulse current method.Two PD models:needle plate defect model and ball plate defect model,are designed.The partial discharge data measured under different aging conditions is studied to evaluate the absolute oil-paper aging state.Aiming at the white noise and periodic narrow-band interference in the PD pulse signal measured by experiment,an improved variational mode decomposition method(VMD)combined with wavelet packet denoising method is proposed in this paper.The effectiveness of this proposed denoising method is verified by the comparative analysis of simulation and measured PD signal.Based on the processed pulse waveform and the measured partial discharge phase distribution spectrum(PRPD),two-dimensional PRPD spectrums are constructed.And the characteristics of PD pulse waveform and the statistical operator characteristics of two-dimensional spectrums are extracted.The three-parameter Weibull distribution model is proposed to describe the distribution law of the occurrence times of different discharge amplitude,and the probability distribution of discharge quantity is obtained by Weibull model.The result shows that the proposed method is effective and the parameters of Weibull model are extracted as PD characteristic parameters.A total of 25 characteristics were extracted,which laid a foundation for using partial discharge to evaluate the aging state of oil-paper insulation.The partial discharge characteristics of aging oil-paper insulation under the different partial discharge models are studied.It is found that with the aging of oil paper insulation under the needle plate model,the number of partial discharge increases significantly,and the phase of discharge has a centralized trend.At the same time,the high amplitude discharge decreases,and is replaced by the low energy discharge with the increasing number.The partial discharge occurred in the positive half cycle of the ball plate model is more concentrated around 90°.The PD occurred in the negative half cycle has two distribution centers in the early stage of aging,while in the late stage of aging,the distribution center of PD occurred in the negative half cycle is between 240°and 2700 and the concentration of the distribution is enhanced;the amplitude of PD in the early stage of aging is small and there are few large discharges,and there are large discharges in both positive and negative half cycles in the late stage.The restricted Boltzmann machine(RBM)combined with gray wolf optimization support vector machine(GWOSVM)algorithm is proposed to build the aging state evaluation model.The RBM network is used to reduce the dimension of feature set,and the GWOSVM algorithm is used to build the evaluation model.According to the actual PD data,the evaluation accuracy of the model in this paper can reach more than 90.48%under different PD models,which can effectively realize the oil detection based on PD Aging state assessment of paper insulation. |
PDF Full Text Request