基于PSO-SVM的高温炉变频电机局部放电预警监测方法  

PSO-SVM Based on Partial Discharge Warning and Monitoring Method for Variable Frequency Motors in High-temperature Furnaces

作　　者：王树彪 秦涛 匡超 赵浩君 卢顺祥 WANG Shubiao;QIN Tao;KUANG Chao;ZHAO Haojun;LU Shunxiang(State Grid Nanjing Power Supply Company,Nanjing 210019,China)

机构地区：[1]国网江苏省电力有限公司南京供电分公司,江苏南京210019

出　　处：《工业加热》2024年第8期76-81,共6页Industrial Heating

摘　　要：在高温炉加热过程中,非均匀的暂态随机状态会影响高温炉变频电机局部放电信号时间特性,导致时间一幅值关系的离散变量不能确定而影响监测效率,为此提出基于PSO-SVM的高温炉变频电机局部放电预警监测方法。首先,采用奇异值分解(SVD)抑制高温炉变频电机局部放电信号中的窄带干扰,获取有用信号;其次,利用局部线性嵌入(LLE)算法提取降维信号中的局部放电特征参数;最后,将其作为输入特征向量构建基于粒子群优化的支持向量机分类函数(PSO-SVM),完成对高温炉变频电机局部放电类型的识别和监测,并以此为依据求出局部放电阈值预警参数和趋势预警缺陷水平,实现对高温炉变频电机局部放电的预警监测。实验结果表明,所提方法能够有效抑制局部放电窄带干扰,并在一定程度上提高了高温炉电机局部放电信号的识别精度和预警监测效率。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 dimensionality reduction signal;Finally,a support vector machine classification function(PSO-SVM)based on particle swarm optimization is constructed 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 efficiency of partial discharge signals in high-temperature furnace motors to a certain extent.

关 键 词：PSO-SVM高温炉变频电机 局部放电预警监测 奇异值分解 窄带干扰抑制 LLE算法 

分 类 号：TM831[电气工程—高电压与绝缘技术]