On-line Detection And Localization Of Partial Discharge Of Cross-linked Polyethylene Cable In Complex Environment

On-line detection and positioning of partial discharge of cable can reflect the insulation status of the cable in time,provide a basis for maintenance,avoid the occurrence of grid operation accidents,and have important significance for ensuring reliable operation of the power supply network.However,due to the complicated working environment of the cable,the partial discharge signal detected by the high-frequency electromagnetic coupling component(HFCT)is seriously interfered by noise.The key to realize the localization of partial discharge is to extract the partial discharge signal from the noisy signal effectively and pick up its first arrival pulse accurately and reliably.This paper studies the partial discharge detection and localization technology of cables in complex noise environments.The transmission characteristics of the partial discharge signal in cable is an important theoretical basis for achieving discharge measurement and reliable location of partial discharge source.The transmission characteristics of the partial discharge signal were analyzed and studied,and the model of cable and its partial discharge detection was built by using PSCAD/EMTDC special software for power analysis.The simulation results are consistent with the theoretical analysis,which verifies the effectiveness of the cable transmission model.White noise is the main interference source in the partial discharge detection signal of the electromagnetic coupling method,which will seriously affect the picking accuracy of the partial discharge signal at arrival time,and even lead to the failure of localization of partial discharge.A two-step time delay estimation algorithm is proposed,which combines the energy mean ratio of long and short windows with time-varying kurtosis and has a certain adaptability to white noise interference.Firstly,the detection and initial positioning of the partial discharge event are realized by the sliding time window by utilizing the characteristics of the sudden change of the signal energy at the time of the partial discharge.Secondly,the peak value of time-varying peak is picked up as the arrival time of partial discharge signal,which has the characteristics of short rise time and the strongest asymmetry.The simulation results show that the algorithm can achieve higher precision wave-to-time picking when the signal-to-noise ratio is higher than-7dB.In order to further enhance the anti-jamming capability of the algorithm,a denoising method based on fast adaptive variational mode decomposition(FAVMD)is proposed.Firstly,the correlation coefficient method is used to adaptively determine the optimal basic mode number,which is applied to the variational mode decomposition,and the partial discharge signal and the noise signal are quickly decomposed into different basic modes to realize the effective separation of the noise signal.Finally,thetime-delay estimation algorithm based on time-varying kurtosis and multi-sensor measurement method are applied to realize the online positioning of the cable partial discharge source.The experimental results show that the algorithm based on FAVMD-time-varying kurtosis has a significantly enhanced ability to adapt to white noise interference.Under the strong white noise environment with signal-to-noise ratio as low as-16 dB,the relative positioning error is only 0.23%,which can meet the localization positioning.Precision requirements.Periodic narrow-band interference is also the main interference source in cable partial discharge detection,but the existing research mainly focuses on white noise suppression,which is not in line with the actual situation.A composite noise suppression method based on FAVMD combined with adaptive wavelet packet decomposition is proposed.Firstly,the periodic narrowband interference,white noise and partial discharge signal are decomposed into different basic modal components by FAVMD.According to the fact that the power of the basic modal component of periodic narrowband interference in DFT power spectrum is much larger than that of partial discharge signal and white noise,the periodic narrowband interference is filtered out and the partial discharge signal with white noise is obtained;then the adaptive wavelet packet decomposition is used to decompose the signal into high-low frequency components,and further extract the partial discharge signal.The simulation results show that the proposed method can effectively suppress the periodic narrowband interference and white noise interference,and better preserve the original features of the partial discharge signal,and have good overall and local similarity.The denoising results are applied to the localization algorithm of multi-sensor partial discharge based on time-varying kurtosis,which can accurately estimate the initial arrival time of the partial discharge.