PSO-SVM Based on Partial Discharge Warning and Monitoring Method for Variable Frequency Motors in High-temperature Furnaces
During the heating process of a high-temperature furnace,the non-uniform transient random state can affect the time characteristics of the partial discharge signal of the high-temperature furnace variable frequency motor,resulting in the uncertainty of the discrete variable of the time amplitude relationship and affecting the monitoring efficiency.Therefore,a PSO-SVM based method for early warning and monitoring of partial discharge of the high-temperature furnace variable frequency motor is proposed.Firstly,singular value decomposition(SVD)is used to suppress narrowband interference in the partial discharge signal of the high-temperature furnace variable frequency motor and obtain useful signals;Secondly,the local linear embedding(LLE)algorithm is used to extract the partial discharge feature parameters from the dimension-ality reduction signal;Finally,a support vector machine classification function(PSO-SVM)based on particle swarm optimization is construc-ted as the input feature vector to identify and monitor the types of partial discharge in high-temperature furnace variable frequency motors.Based on this,the partial discharge threshold warning parameters and trend warning defect levels are calculated to achieve early warning and monitoring of partial discharge in high-temperature furnace variable frequency motors.The experimental results show that the proposed method can effectively suppress narrowband interference of partial discharge and improve the recognition accuracy and early warning monitoring effi-ciency of partial discharge signals in high-temperature furnace motors to a certain extent.
PSO-SVMhigh temperature furnace variable frequency motorpartial discharge warning and monitoringsingular value decom-positionnarrowband interference suppressionLLE algorithm
NETL
NSTL
万方数据
