摘要
为了提取机械设备故障引发的冲击成分,提出了一种基于非线性流形学习的冲击故障特征自适应提取方法.该方法将反映故障的振动信号重构到高维相空间中,利用局部切空间排列的流形学习方法提取出隐藏其中的低维流形,并基于峭度和偏斜度指标的特点,提出了冲击波形量化的取值策略,实现了高维相空间中局部邻域参数的自适应选取,从而提取出最优的冲击故障特征.通过仿真数据的对比分析和工程应用,表明该方法能够较好地提取出冲击成分信号,与小波软阈值方法相比,提取出的冲击特征成分更完整,周期性更好.
To acquire the impact component aroused by mechanical fault, a new feature extraction method based on manifold learning is proposed. After embedding the raw vibration signal into a high dimensional phase space to reconstruct a dynamical manifold, the local target space alignment algorithm is employed for extracting nonlinear low dimensional manifold. According to the characteristics of the kurtosis index and skewness index, the adaptive selection criterion of local neighborhood parameters in phase space is introduced to reflect the optimal impacts. The experimental results and industrial measurements show that this approach, compared with the softthreshold method, is more effective to extract the weak periodic impacts from mechanical signals.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2009年第11期95-99,共5页
Journal of Xi'an Jiaotong University
基金
科技部国家"863计划"资助项目(2007AA04Z432)
关键词
流形学习
特征提取
冲击故障
manifold learning
feature extraction
impact response
