Research On Partial Discharge Data Processing Technology Of High Voltage Motor

The reliable operation of high-voltage motors is essential to maintain the production and safety of industrial facilities.For a long time,the condition monitoring of high-voltage motors has relied on offline testing.Because online monitoring technology can provide more effective equipment status information,it has been developed rapidly in the past ten years,and a lot of research has been conducted.Especially in the oil and gas industry,the unplanned shutdown of equipment has a great impact on the normal production of enterprises.Operation and maintenance personnel can easily understand the insulation status of the motor through the identification and analysis of the partial discharge signal of the high-voltage motor.It is of great significance to further determine the equipment maintenance interval and avoid unplanned shutdowns.Based on the analysis of the relationship between the insulation structure of the high-voltage stator windings and the partial discharge of the windings,and sorting out the types of partial discharges of the stator windings,the problem of complicated motor operating conditions and the partial discharge signal being often obscured by noise during the on-line monitoring of partial discharges is solved.Based on the research on denoising technology of existing signal data,the existing variational modal decomposition algorithm is improved to make it more suitable for denoising of partial discharge signals.Aiming at the problem of difficulty in separating multi-source partial discharge signals of stator windings when the motor is running,based on the research of multiple data separation technologies,the density peak clustering algorithm is optimized,and the density peak clustering algorithm is used to process high-dimensional data.Difficulty This difficulty is optimized using principal component analysis dimensionality reduction algorithm.Correctly extracting the features of the original signal is the prerequisite for clustering,which has a greater impact on the clustering results.Based on the investigation of various signal feature extraction algorithms,wavelet coefficients are used to characterize the characteristics of local signals,and the partial discharge signal multi-source separation is finally realized through signal feature extraction,dimensionality reduction and clustering of local signals.Based on the research on the two data processing algorithms of denoising and multi-source separation,a data processing system for partial discharge monitoring of high-voltage motors has been developed,which can realize single-signal spectrum analysis,denoising,amplitude analysis and other functions.The partial discharge signal can carry out multi-source separation.The improved variational modal decomposition denoising algorithm can obtain partial discharge signals with higher signal-to-noise ratio,and the improved peak density clustering algorithm is used to separate partial discharge data from multiple sources with high efficiency and accuracy.In this paper,a partial discharge data processing system for high-voltage motors is developed on the basis of the aforementioned data processing algorithm.Many researches in this article provide a basis for correctly understanding the insulation state of high-voltage motors and then determining the equipment maintenance plan,and are a powerful guarantee to avoid unplanned shutdowns of high-voltage motors.