摘要
滚动轴承在发生严重故障前会经历不同的退化状态,针对时域和频域故障特征不能表征早期故障的问题,提出了小波包能量结合高斯混合模型的轴承性能退化指标提取方法。该方法以小波包能量比例值向量作为原始特征,引入高斯混合模型,以计算出的小波包能量比例值对数似然概率作为性能退化指标,通过实验验证该方法能发现早期故障,还能很好的跟踪轴承退化趋势。在此基础上,由于退化状态难以界定识别,利用模糊C均值聚类对性能退化指标模糊聚类,从而识别轴承性能退化状态,通过滚动轴承退化实验验证了该方法的有效性。
Rolling bearing Undergo different degradation state before happens serious breakdown, aimed at the problem of Time domain and frequency domain fault feature can not represent early fault. A method for Bearing performance degradation index extraction of wavelet packet energy combined with Gauss's hybrid model is proposed. In this method, the wavelet packet energy ratio value vector is used as the original feature, introduce Gauss mixture model, take Wavelet packet energy ratio5 sLog likelihood probability as performance degradation index, The method can detect the early faults and can track the degradation trend of the bearing, on this basis, Due to the ambiguity of different degradation States, Fuzzy clustering Degradation index by using Fuzzy Center Means, Thereby identifying the bearing performance degradation state, Through the rolling bearing degradation experiments verify the validity of the method.
出处
《组合机床与自动化加工技术》
北大核心
2016年第11期92-95,共4页
Modular Machine Tool & Automatic Manufacturing Technique
关键词
滚动轴承
小波包变换
特征提取
模糊聚类
rolling bearing
wavelet packet transform
feature extraction
fuzzy clustering