基于振动信号EMD分解与支持向量机的有载分接开关机械故障诊断

Mechanical Fault Diagnosis of On-load Tap-changer Based on EMD Decomposition of Vibration Signal and Support Vector Machine

本文中作者将经验模态分解与支持向量机算法引入到有载分接开关机械故障识别领域。通过对有载分接开关振动信号进行经验模态分解,并选取前5个本征模态函数的能量作为输入特征参量,利用支持向量机算法对有载分接开关弹簧储能不足、触头松动、触头磨损3种机械故障进行了识别,同时比较了不同核函数对支持向量机分类结果的影响。

Empirical mode decomposition and supported vector machine is introduced in fault diagnosis of on-load tap changers(OLTC).By making empirical mode decomposition of vibration signal of OLTC,and making the first five intrinsic mode function(IMF)as the input characteristic parameters of supported vector machine,three faults of OLTC including tap abrasion,tap loosening,and insufficient energy-storage are recognized.The influence of different kernel functions is discussed.