摘要
滚动轴承性能退化评估是实现轴承剩余寿命预测和制定维修方案的基础,聚类分析作为数据挖掘的基本工具在故障评估中得到广泛应用。K-medoids聚类算法不易受极端数据影响,与隶属度函数相结合计算隶属度值作为故障指标可将故障程度大小表示在特定区间上,Renyi熵作为特征能很好地识别信号的微小变化。对滚动轴承振动信号进行小波包分解并降噪,将最优子带进行重构后计算Renyi熵,构成特征向量输入到K-medoids聚类模型中得到隶属度作为评价指标评估当前轴承性能状态,实现对故障的定量评估,同时设定自适应阈值确定早期故障出现时间。人工植入故障试验和全寿命疲劳试验分析表明该方法能有效评估故障程度大小且能及时发现早期故障的出现。
Bearing performance degradation assessment(PDA)underlies residual useful life prediction and maintenance decision-making.In bearing fault diagnosis and PDA,a commonly used method is clustering analysis,among which the K-medoids clustering is not susceptible to extreme data.The amalgamation of the K-medoids clustering and a membership function are expected to give a health indicator with a determined value range for PDA.Vibration signals are first decomposed into multiple sub-bands on which a denoising algorithm is then applied,resulting in a series of denoised sub-band signals.The Renyi entropy is a nonlinear parameter able to identify subtle change in signals.As such,Renyi entropies of denoised sub-band signals constitute the feature vectors presented to the combined model of K-medoids based clustering and a membership function.Resultant health indicator was used to depict bearing health condition with the help of an adaptive threshold.Results on artificially induced faults and bearing run-to-failure data demonstrate that the proposed method is able to track the progress of bearing faults and detect them at incipient stage.
作者
张龙
宋成洋
邹友军
洪闯
王朝兵
ZHANG Long;SONG Chengyang;ZOU Youjun;HONG Chuang;WANG Chaobing(East China Jiaotong University,School of Mechatronics&Vehicle Engineering,Nanchang 330013,China;CRRC,Qishuyan CO.,LTD.,Changzhou 213011,China)
出处
《振动与冲击》
EI
CSCD
北大核心
2020年第20期24-31,46,共9页
Journal of Vibration and Shock
基金
国家自然科学基金(51665013,51865010)
江西省自然科学基金(20161BAB216134,20171BAB206028,20171BAB216030)。
作者简介
第一作者:张龙,男,博士,副教授,1980年生;通信作者:汤文辉,男,博士,教授。
