故障诊断中的多重分形分析

Multi-fractal analysis on fault diagnoses

建立了模拟实际故障的实验装置,采集了碰摩故障、油膜振荡及碰摩和油膜振荡相互作用的耦合故障振动信号。运用多重分形理论,对实测故障信号计算出多重分形广义维数,重点讨论广义分形维数谱和奇异谱,并提出广义维数谱能和奇异谱能的概念,并以两谱能作为特征量,实现对故障特征的提取与识别。研究表明:将广义维数谱能和奇异谱能结合使用,有利于分析识别故障信号,增强可靠性。该研究为复杂旋转机械故障诊断提供了一种识别方法。

The experimental equipment is established and the fault vibration signals are collected which are the rub-impact fault,oil oscillating and rub and oil oscillating interaction coupling fault using multi-fractal theory,general fractal dimension of fault signals is calculated,and general dimension spectrum and singular spectrum are discussed deeply.Hence general dimension energy spectrum singular energy spectrum is defined,and two energy spectrums are used as characteristic quantity to actualize the feature extraction and recognition.Study shows that fault signals are identify effectively and enhance robustness with general dimension energy spectrum and singular energy spectrum used together.It provided an identification method for the complex rotating machinery fault diagnosis.