形态学多尺度广义分形矩阵在轴承故障诊断中的应用  

Application of Morphological Multi-Scale Generalized Fractal Matrix in Fault Diagnosis for Bearings

作　　者：于忠德[1] 刘东升[2] 王锋[2] 李兵[2] 

机构地区：[1]沈阳职业技术学院,沈阳110000 [2]军械工程学院四系,石家庄050003

出　　处：《轴承》2017年第7期48-51,共4页Bearing

摘　　要：针对轴承故障信号非线性特征提取的问题,提出一种形态学多尺度广义分形矩阵对轴承故障信号进行表征。形态学多尺度广义分形矩阵采用数学形态学覆盖法估计信号的分形维数,同时考虑信号在尺度和统计分布上的非严格自相似性,包含比传统分形维数更为丰富和全面的信息。采用轴承7种状态下的振动信号对形态学多尺度广义分形矩阵进行验证,结果表明:与传统的单一分形维数、广义分形维数和多尺度分形维数相比,形态学多尺度广义分形矩阵具有更高的轴承故障诊断精度。Aiming at nonlinear feature extraction of fault signals for bearings, a morphological multi - scale generalized fractal matrix is proposed to characterize fault signals for bearings. The mathematical morphology covering method is used by morphological multi - scale generalized fractal matrix to evaluate fractal dimension of signals, and the non - strict self- similarity of signals on scale and statistical distribution is taken into account, which containing more plentiful and comprehensive information than traditional fractal dimension. The vibration signals of the bearings under seven states are used to verify morphological multi - scale generalized fractal matrix. The results demonstrate that the morpho- logical multi -scale generalized fractal matrix gives a better fault diagnosis accuracy for the bearings than traditional single fractal dimension, generalized fractal dimension and multi -scale fractal dimension.

关 键 词：滚动轴承 故障诊断 数学形态学 多尺度广义分形矩阵 特征提取 

分 类 号：TH133.33[机械工程—机械制造及自动化]