摘要
根据不同故障类型的轴承信号在高维相空间中呈现不同结构的流形形态,提出了基于流形拓扑结构的轴承故障无监督分类方法。新方法首先将反映轴承状态的一维振动信号重构到高维相空间中,利用相点邻域的切空间信息逼近流形的局部几何结构,从而得到描述流形拓扑结构的矩阵;对所有样本构成的流形拓扑结构作多向主元分析后,将获得的主元信息作为特征集输入到C-均值分类器中进行轴承的状态识别。用轴承在正常状态、内圈故障、外圈故障的试验数据进行验证,结果表明,与传统的利用振动统计量为特征输入的方法相比,新方法能够更完整地刻画信号特征,获得更准确的分类识别率。
An unsupervised classification method is proposed according to the fact that the signals of different bearing faults have different topological structures of manifolds. After embedding the time series into a high dimensional phase space to reconstruct a dynamical manifold, the topological structure containing the information of the manifold evolution is acquired by extracting the local geometric information of each point with its neighborhood data. The largest principal components of the overall manifold structure are calculated by performing multi-way principal component analysis to the topological structures matrices, and are input as features into C-means classifier, which output the classification results. Taking roller bearing vibration signals of normal state, outside-rolling and inside-rolling faults as samples, the experimental results show that the proposed method has better classification performance than the traditional method.
出处
《控制工程》
CSCD
北大核心
2009年第3期358-362,共5页
Control Engineering of China
基金
国家自然科学基金资助项目(11240035)
机械传动与制造工程湖北省重点实验室基金资助项目(3062012)
关键词
无监督分类
故障诊断
流形拓扑结构
多向主元
unsupervised classification
fault diagnosis
topology structure of manifold
multi-way principle component