基于振动信号样本熵和相关向量机的万能式断路器分合闸故障诊断

发布时间：2018-10-20 17:42

【摘要】：为实现对万能式断路器分合闸故障的非侵入式监测和诊断,以分合闸过程中所产生的包含丰富机械特性信息的振动作为信号来源,提出一种基于振动信号互补总体平均经验模态分解(CEEMD)-样本熵和相关向量机(RVM)相结合的万能式断路器故障诊断方法。该方法首先将振动信号通过改进的小波包阈值去噪算法处理;其次采用CEEMD提取若干个反映断路器状态信息的固有模态函数(IMF)分量,依据各IMF分量的能量分布特点,选择其中前7阶进行处理,计算其样本熵形成有效的特征样本;最后通过计算不同故障类型的样本间欧氏距离来定量评价类间样本平均距离,建立基于RVM的二叉树多分类器,诊断得出万能式断路器故障类型。基于所设计的分合闸典型故障模型进行实验。与其他方法的对比实验表明,所提方法可利用相对较少的故障数据样本实现对万能式断路器故障类型的识别并具有较高的识别率;同时实验表明,辅以同一故障类型的样本间欧氏距离,可实现对分合闸故障中三相不同期故障严重程度的初步评估。
[Abstract]:In order to realize the non-invasive monitoring and diagnosis of commutative circuit breakers, the vibration, which contains abundant information of mechanical characteristics, is used as the signal source. A fault diagnosis method for universal circuit breakers based on vibration signal complementary population average empirical mode decomposition (CEEMD)-sample entropy and correlation vector machine (RVM) is proposed. Firstly, the vibration signal is processed by the improved wavelet packet threshold de-noising algorithm; secondly, the (IMF) components of several inherent mode functions reflecting the state information of circuit breakers are extracted by CEEMD, and the energy distribution characteristics of each IMF component are analyzed. The first seven orders are selected for processing, and the entropy of the sample is calculated to form an effective feature sample. Finally, by calculating the Euclidean distance between samples of different fault types to quantitatively evaluate the average distance between the samples, a binary tree multiclassifier based on RVM is established. The fault type of universal circuit breaker is obtained by diagnosis. Experiments are carried out based on the designed fault model. The comparison experiments with other methods show that the proposed method can recognize the fault type of the circuit breaker with relatively small number of fault data samples and has a high recognition rate. The Euclidean distance between samples of the same fault type can be used to evaluate the severity of three-phase faults at different stages.
【作者单位】：河北工业大学控制科学与工程学院;电磁场与电器可靠性省部共建重点实验室(河北工业大学);
【基金】：河北省教育厅资助科研项目(ZD2016108)
【分类号】：TM561

【相似文献】