摘要
振动频谱类型的识别决定汽轮机振动故障的诊断方向。为了提高在线识别精度,创新性地将核独立元分析(KICA)和故障重构思想引入振动频谱识别,提出一种基于振源方向的频谱类型识别方法。首先,基于核独立元分析方法,从频谱数据中提取激振力信息。然后,借鉴故障重构的思想,在样本独立元空间中提取故障振源特征方向,建立频谱类型识别库。最后,通过实时对振动频谱在各类故障特征方向上的重构识别,判断频谱类型。采用该方法对试验台数据和某机组实际振动频谱数据的识别进行了验证,并与基于欧氏距离的相似性识别方法进行了对比,证明了该方法的优越性。基于提取振源方向的频谱类型识别的研究,可为汽轮机振动故障智能诊断提供更加准确、可靠的频谱征兆信息,也为同类机械振动频谱智能识别提供了新的解决方案。
The identification of vibration spectrum type determines the diagnosis direction of turbine vibration fault.In order to improve the accuracy of on-line identification of vibration spectrum,innovatively,an idea of kernel independent component analysis(KICA) and fault reconstruction is introduced into vibration spectrum identification,and a new identification method based on the direction of vibration source is proposed.Firstly,the excitation force information is extracted from high-dimensional spectrum data by means of KICA.Then,based on the idea of fault reconstruction,the feature direction of fault vibration source is extracted from the independent component space,and the spectrum type recognition library is established.Finally,the spectrum types are identified by real-time reconstruction on each feature direction of various kinds of fault.The method is validated in the identification of the experimental data and the actual vibration spectrum data of a certain turbine,and compared with the similarity recognition method based on Euclidean distance,which proves the superiority of the method.Based on the research of the spectrum type identification of the direction of vibration source,it can provide more accurate and reliable symptom information for the intelligent diagnosis of turbine vibration fault,and also provide a new solution for the intelligent identification of similar mechanical vibration spectrum.
作者
顾煜炯
杨楠
孙树民
刘璐
徐教辉
GU Yujiong;YANG Nan;SUN Shumin;LIU Lu;XU Jiaohui(School of Energy,Power and Mechanical Engineering,North China Electric Power University,Beijing 102206,China;School of Control and Computer Engineering,North China Electric Power University,Beijing 102206,China)
出处
《自动化仪表》
CAS
2019年第2期1-7,共7页
Process Automation Instrumentation
基金
国家重点研发计划基金资助项目(2017YFB0603904-4)
中央高校基本科研业务费专项基金资助项目(2016XS35
2016MS44)
关键词
汽轮机
核独立元分析
故障重构
频谱识别
智能诊断
数据挖掘
Turbine
Kernel independent component analysis(KICA)
Fault reconstruction
Spectrum identification
Intelligent diagnosis
Data mining