摘要
在对信号设备的故障诊断中,训练集在检测时对核密度估算不足,导致诊断方法在进行全失稀疏编码特征提取时,故障特征信号不明显,容易出现漏诊。因此,提出基于稀疏残差距离的铁道信号设备故障诊断方法。首先对接收到的设备信号使用过完备字典进行稀疏化,将稀疏化后的信号作为测量样本,使用欧式距离法检测信号核密度,并得出故障判断,在确定出现故障后,使用故障树法对故障信号进行重要度割集,完成对设备故障的诊断。为了验证设计的设备故障诊断方法的有效性,设计实验,使用设计方法和全失NA-MEMD方法、全失EEMD方法、文献[2]中的故障诊断方法对出现故障的MCP23017铁道信号机进行诊断,实验结果显示,设计的方法对故障设备的全失稀疏编码特征提取较好,故障特征信号明显,证明设计方法的有效性。
in the fault diagnosis of signal equipment,the kernel density is not estimated enough in the training set,which leads to the fault feature signal is not obvious and easy to be missed when the diagnosis method carries out the feature extraction of total loss sparse coding.Therefore,a railway signal equipment fault diagnosis method based on sparse residual distance is proposed.Firstly,the received equipment signals are sparse using over complete dictionary.The sparse signals are taken as measurement samples,and the kernel density of signals is detected by Euclidean distance method,and the fault judgment is obtained.After the fault is determined,the fault tree method is used to cut the importance of fault signals to complete the fault diagnosis.In order to verify the effectiveness of the designed equipment fault diagnosis method,the design method,total loss na-memd method,total loss EEMD method and fault diagnosis method in reference[2]are used in the design experiment to diagnose the fault mcp23017 railway signal.The experimental results show that the design method can extract the full loss sparse coding features of fault equipment well,and the fault characteristic signal is obvious.The design method is effective.
作者
李晓艳
于明
LI Xiaoyan;YU Ming(Shanxi Railway Institute,Weinan Shanxi 714000,China)
出处
《自动化与仪器仪表》
2021年第1期51-54,共4页
Automation & Instrumentation
基金
陕西省教育厅高等教育教学改革研究项目,高速铁路综合维修技术专业群(工电供专业融合)构建与实践(No.19GY025)。
关键词
信号设备
核密度
故障诊断
过完备字典
重要度割集
signal equipment
kernel density
fault diagnosis
over complete dictionary
importance cut set
