Optimization Of Adaptive Noise Reduction Method For High Frequency Partial Discharge Signal Of Vehicle-mounted Cable Terminal

As an important hub of electrical connection between the external pantograph and the side cabinet of the high voltage network inside the locomotive,the running state of the on-board cable directly affects the safety and stability coefficient of rail transit.The cable terminal is the core part of the on-board cable,its insulation structure is relatively weak,more prone to failure,therefore,the real-time monitoring of the on-board cable terminal insulation state is very crucial.The high frequency partial discharge test system is an important means to detect the insulation state of the cable terminal,which can diagnose the cable terminal fault in time.The detection signal generated by partial discharge is very weak.In terms of the value,it is easy to be submerged by the external interference signal,especially the complex environment interference in the field condition of the train has a particularly serious impact on the detection results,which often leads to the weak high frequency partial discharge signal in the time domain is completely annihilated by the strong interference signal in the time domain,so it is hard to recognize whether the partial discharge occurs,which brings great difficulties to the characteristic analysis of partial discharge signal in the later period and the evaluation of aging state.Therefore,it is very important to realize effective interference suppression in the detection of high-frequency partial discharge signal at the on-board cable terminal.In this paper,the field environmental interference and the corresponding interference suppression technology are studied in the test of the high frequency partial discharge signal of the on-board cable terminal.In view of the possible interference in the field high frequency partial discharge test of the on-board cable terminal,the analysis is made from the interference signal source and the interference propagation path,and the final test condition is selected under the maintenance state after the train enters the station.At this time,the main interference is white noise and narrowband interference signal.For narrowband interference,narrowband interference suppression method based on generalized S transform and singular value decomposition is optimized.In this method,the number of narrowband interference is obtained by the generalized S-transformation adaptively.Then the improved 3 spectral line interpolation method is proposed to accurately estimate the narrowband interference frequency and construct the guide signal to be added to the original signal.This signal is decomposed by the singular value to suppress the periodic narrow-band interference.Aiming at white noise interference,a white noise suppression method based on empirical mode decomposition and wavelet packet transform is improved.The method firstly intercepts high frequency partial discharge signal fragments by means of generalized S transformation combined with curve threshold.The extreme value envelope curve was obtained after the extension of both ends of each partial discharge signal fragment,and empirical mode decomposition was performed after the truncation of the envelope in accordance with the original segment,so as to effectively eliminate the influence of end point effect.Combining with the characteristics of each partial discharge signal segment,high frequency harmonic signals are introduced into the original signal segment for empirical mode decomposition.Combined with the correlation number to screen out useful signal IMF component and combined with wavelet packet threshold denoising,the partial discharge signal with white noise suppression can be obtained.The proposed method is compared with the traditional noise reduction method by combining the simulation signal with noise and the measured high frequency partial discharge signal.The results show that,the narrow-band interference suppression method optimized in this paper has better narrow-band interference suppression effect.At the same time,the method can effectively avoid the influence of frequency aliasing,and the reduction degree of high-frequency partial signal is higher.In addition,the white noise reduction method proposed in this paper is optimized on the basis of empirical mode decomposition,and the dominant component of the noise can be directly filtered out.This method has a more significant effect on the noise reduction of the high frequency partial discharge signals,especially the measured signals.In conclusion,the two optimization methods in this paper have better noise reduction effect than the traditional noise reduction method,and the reduction degree of high frequency partial discharge signal is higher,which has a good application prospect.