Partial Discharge Signal Denoising And State Recognition Analysis Of Oil-paper Insulation Heat Aging

As the key equipment of the power system,the transformer bears significant operational responsibility,so its insulation state is especially important for the safe and stable operation of the system.Oil-paper insulation is the main form of internal insulation of transformers,and the overall reliability of the transformer insulation performance depends on it.As a non-destructive testing method,partial discharge is widely used in the evaluation of transformer oil-paper insulation aging state.However,partial discharge works in a strong electromagnetic environment and is highly random.At the same time,there is a nonlinear correspondence between the feature quantity and the aging stage.Based on a variety of characteristic parameters,this paper uses the modified quantum particle swarm optimization support vector machine algorithm to evaluate the aging state,thereby further improving the accuracy of the evaluation of the aging state of transformer oil-paper insulation.In this paper,a 21-day accelerated thermal aging experiment was carried out on kraft paper and Nomex paper in the laboratory.The partial discharge signal of insulating paper in different aging stages was measured.The partial discharge signal was denoised,extracted and characterized,and the aging state of the insulating paper was finally identified.The paper mainly includes the following contents:(1)A new method of partial discharge denoising based on 2FFT improved sparse decomposition maching pursuit algorithm(MP-2FFT)algorithm is proposed for nonlinear and non-stationary characteristics of partial discharge signals.The MP-2FFT algorithm is used to denoise the partial discharge simulation signal,and compared with the traditional wavelet threshold method and the modal decomposition method.The results showed that the denoising effect of the MP-2 FFT algorithm is superior to the traditional algorithms in all evaluation indexes.Compared with the MP algorithm,the convergence rate of MP-2FFT algorithm is significantly faster.At the same time,the MP-2 FFT algorithm is used to denoise the measured partial discharge signal.(2)The characteristics of the partial discharge time series is constructed.The time series of partial discharge signals in different aging stages of kraft paper and Nomex paper were used to extract statistical characteristic parameters,chaotic characteristic parameters and improve the Shannon entropy.The variation rules of the characteristic parameters of the two insulating papers in different aging stages were compared and analyzed,and the time-sequence characteristics of the partial discharge mixing of the insulating paper with heat aging were obtained.(3)The modified quantum particle swarm optimization inertia factor is corrected to optimize the support vector machine for oil film aging stage identification.Correcting the inertia factor variation formula improves the convergence speed of the algorithm and ensures the accuracy of the quantum particle swarm optimization algorithm.The modified quantum particle swarm optimization inertia factor is corrected to optimize the support vector machine algorithm is used to identify the aging phase of the insulation paper effectively based on the basic partial discharge characteristic parameters and their combinations.The hybrid time series feature can further improve the recognition accuracy of the modified quantum particle swarm optimization inertia factor is corrected to optimize the support vector machine algorithm on the aging state of oil-paper insulation.