摘要
为从齿轮振动信号中提取包含故障信息的特征量,提出了一种基于经验模态分解(EMD)降噪的递归图分析方法。该方法用EMD方法将振动信号分解为有限个固有模式函数(IMF)分量,选取包含故障信息的IMF分量建立递归图,从递归图中提取特征向量,运用高斯混合模型进行模式识别。将该方法运用于故障齿轮振动信号的识别,结果表明该方法具有较高的识别率,对齿轮故障能够有效地进行分类与诊断。
In order to extract the characteristic quantity which contained failure information from the gear vibration signals,a recurrence plot analysis method based on EMD was proposed. The vibration signal was decomposed into a finite number of IMF using EMD method. Recurrence plot was established by the IMF which contained failure information. Characteristic quantities were extracted from the recurrence plot. The patterns were identified by gaussian mixture model. The results show that the method has a higher recognition rate,it's effective to the gear failure's classification and diagnosis.
出处
《机床与液压》
北大核心
2015年第5期160-163,共4页
Machine Tool & Hydraulics
基金
国家自然科学基金青年基金资助项目(51105284)
