摘要
特征数据集的属性约简是机械故障智能诊断的关键步骤之一。目前利用粗糙集理论从大量的且含有噪声以及非线性、非平稳信号的故障数据集中提取出有用特征信息是一件值得研究的事情。针对原始故障数据集直接离散化会导致一些关键属性丢失以及时域内分析不能有效获取故障本质的问题,提出了一种以频域内的频谱值为条件属性,以故障类别为决策属性建立邻域粗糙集决策表对数据集进行属性约简的方法。通过处理转子实验台数据对该方法进行验证和对比,结果表明该方法能有效地获得典型故障的关键属性和更加准确的决策规则。
The attribute reduction of the characteristic data set is one of the key steps in mechanical fault intelligent diagnosis.It is something worth researching to select the useful features from the high-dimensional nonlinear and non-stable data set mixing noise and fault based on rough set theory(RST)currently.However,dispersing the original data set directly leads to loss of some important attributes,and getting the nature of fault is not effective in time domain.Thereby,this paper uses frequency spectrum as condition attributes and faults as decision attributes,establishes a neighborhood rough set model to process attribute reduction.Experimental results show that the proposed method can effectively obtain key attributes of different typical faults and decision rule.
作者
何敬举
赵荣珍
赵孝礼
孙业北
HE Jingju;ZHAO Rongzhen;ZHAO Xiaoli;SUN Yebei(College of Electrical and Information Engineering, Lanzhou University of Technology,Gansu Lanzhou,730050,China)
出处
《机械设计与制造工程》
2018年第3期22-26,共5页
Machine Design and Manufacturing Engineering
基金
国家自然科学基金资助项目(51165019)
关键词
属性约简
故障诊断
频谱
粗糙集
邻域粗糙集
attribute reduction
fault diagnosis
frequency spectrum
rough set
neighborhood rough set
