基于VMD-MSE的高压断路器故障诊断

Fault diagnosis method of high-voltage circuit breaker based on VMD and MSE

为了有效提取高压断路器振动信号的特征,提出了一种基于变分模态分解(VMD)和多尺度熵(MSE)的特征提取方法.首先,利用VMD对断路器的振动信号进行分解,得到一系列的固有模态分量;然后,求解各模态分量的MSE值,将得到的熵值进行组合作为初始特征向量,再应用主成分分析法(PCA),提取累计贡献率达到85%的特征主成分作为断路器振动信号的特征向量;最后,采用支持向量机(SVM)对4种运行状态下的断路器进行模式识别.实验结果表明:该方法能够有效提取断路器振动信号的特征,以其作特征向量能够较好地判断出断路器的故障类型,对于断路器的故障诊断和检修维护具有重要的意义.

In order to extract the feature of the vibration signals from high-voltage circuit breaker effectively,a feature extraction method based on variation mode decomposition (VMD) and multi-scale entropy (MSE) is proposed.Firstly,the vibration signal of circuit breaker is decomposed by VMD and a series of intrinsic mode function (IMFs) is obtained.Then,the MSE value of each IMFs is calculated and combined as initial feature vector,and then,the major ingredient of the initial feature vector is decomposed by principal component analysis (PCA) as the feature vector of circuit breaker,whose accumulative contribution rate reaches 85%.Finally,support vector machine (SVM) is used to identify the fault types of circuit breaker under four working conditions.The experimental results show that:the proposed method can effectively extract the feature of vibration signals from high-voltage circuit breaker,and the feature vector can determine the fault type of circuit breaker well,which is of great significance for fault diagnosis and maintenance of circuit breaker.