基于VMD和样本熵的高压断路器故障特征提取及分类

Fault feature extraction and classification of high voltage circuit breakers based on VMD and sample entropy

提出了一种基于变分模态分解(Variational Mode Decomposition,VMD)和样本熵的高压断路器振动信号的特征向量提取方法,并采用支持向量机(Support Vector Machine,SVM)对故障类型进行识别。将断路器振动信号进行滤波处理,对信号进行变分模态分解,利用分解得到的固有模态函数分量(Intrinsic Mode Function,IMF)表征断路器各个振动事件,计算其样本熵作为特征向量,利用SVM对断路器不同运行状态进行分类识别。仿真信号表明,VMD对于处理瞬态非周期性的振动信号具有优越的分解特性。利用该方法在实验室条件下对四类故障状态进行特征提取和识别,对比结果表明应用该方法能有效提取高压断路器的故障特征并准确地识别出故障类型。

A new method for fault feature extraction of high voltage circuit breakers was proposed based on variation mode decomposition(VMD)and sample entropy,and the support vector machine(SVM)was utilized to recognize the fault types.Firstly,after the vibration signal of the circuit breakers s preprocessed,the signal was decomposed by VMD.The decomposed IMFs were used to characterize various vibration events of the circuit breakers.Then the feature vector was acquired by calculating the sample entropy of IMFs.Finally,SVM was used to classify different operating states of circuit breakers.Simulation signals show that VMD has superior decomposition characteristics to deal with transient aperiodic vibration signals.Four types of fault states were extracted and recognized using the above method.The comparison results show that this method can extract fault characteristics effectively and classify fault types of high voltage circuit breaker accurately.