针对现有研究未充分关注控制棒驱动机构(control rod drive mechanism,CRDM)的早期故障诊断问题、很难将故障特征定位至具体部件以及人工引入的故障样本与装备实际故障特征存在差异等不足,提出了一种基于振动信号的CRDM滚轮早期故障诊断方法:首先,利用寿命考核试验时机采集了某密封磁阻马达式CRDM的滚轮全寿命振动信号,基于经验模态分解(empirical mode decomposition,EMD)和Hilbert变换方法进行解调分析,获得与滚轮退化状态相关的模态成分;然后,采用时、频域分析方法获得了 11个能够直接表征CRDM滚轮磨损状态的特征量,并根据退化趋势提取出与实际故障特征高度吻合的早期故障样本;最后,分别基于BP神经网络和支持向量机两种方法实现了 CRDM滚轮早期故障的多特征智能诊断.结果表明:提取的滚轮早期磨损故障样本与实际运行过程保持了较好的一致性,证明所提CRDM滚轮早期故障诊断方法具有较强的工程应用价值.
Incipient fault diagnosis of control rod drive mechanism roller based on vibration signal
In response to the insufficient attention paid by existing research to the incipient faults diag-nosis of control rod drive mechanism(CRDM),the difficulty in locating faults features to specific components,and the differences between manually introduced fault samples and actual equipment fault features,a vibration signal based incipient faults diagnosis method for CRDM roller was pro-posed.The roller's full life vibration signals of a sealed reluctance motor type CRDM was collected during the life assessment experiment.Demodulation analysis was conducted based on EMD and Hil-bert transformation methods and modal components related to the degradation state of the roller were obtained.Through time and frequency domain analysis methods,11 feature parameters that could di-rectly characterize the wear status of CRDM roller were obtained,and incipient fault samples with high consistency to actual fault features were extracted based on the degradation trend.Finally,multi-ple feature intelligent diagnosis of incipient faults of CRDM roller was achieved based on BP neural network and SVM methods.The early wear fault samples of the roller extracted by the proposed method are consistent with the actual operating process,which proves high engineering application value of the proposed incipient fault diagnosis method of CRDM.
vibration signalcontrol rod drive mechanismincipient faults diagnosisdemodulation analysis
NETL
NSTL
万方数据
