滚动轴承振动信号的非线性分形特性研究

摘要

为研究球轴承振动信号的非线性分形特性，对 Hurst 指数及多重分形计算分形特征的方法进行改进。采用2次分割充分利用数据，并对不同状态振动的 Hurst 指数和多重分形谱的变化规律进行了仿真及试验研究。结果表明：正常状态振动呈现正的持续性，故障状态振动具有反持续性，故障状态振动的波动影响范围比正常状态振动的小；振动信号具有不均匀的自相似性（即多重分形特征），振动状态与多重分形谱的形状之间存在着对应关系。%In order to study nonlinear fractal characteristics of vibration signals on ball bearings,the computational method for Hurst index and multifractal are improved.The data is efficiently used by double segmentation,the variation rule for Hurst index and mulifractal spectrum of vibration under different states are simulated and experimentally stud-ied.The results show that the vibration under normal state is correlated positively,the vibration under abnormal state is correlated negatively,but the influence of normal state on signal in fluctuations is stronger than that of abnormal states;the vibration signal have characteristics of an uneven self -similarity (e.g.multifractal behavior),there is a corre-sponding relationship between vibration state and shape of mulifractal spectrum.