摘要
以故障信号局部包含信息的差异性为基础,通过短时奇异值分解算法来提取淹没与背景噪声中的故障冲击成分。再结合K-SVD稀疏分解训练算法,提出1种自适应学习字典构建方法,可以有效的自适应表征滚动轴承的故障信号。仿真和试验结果分析论证表明,该故障特征提取技术较常规稀疏匹配算法具更好的识别和提取冲击故障特征能力,有助于实现滚动轴承故障智能诊断。
Aiming to extract the fault impulse feature information from strong background noise efficiently,a slipping singular value decomposition based on the difference in local feature of fault signal is introduced to fault diagnosis for rolling bearings. Combined with K-SVD sparse decomposition algorithm,a new adaptive learning dictionary method is put forward. And the proposed method is effectively and successfully applied to the weak fault diagnosis. Some simulation and experiment results validate that the proposed method shows better ability to identify and extract weak shock fault feature information compared with other traditional sparse decomposition methods,which is greatly help to the intelligent diagnosis of rolling bearings.
作者
唐宁
童水光
徐剑
从飞云
张依东
TANG Ning;TONG Shuiguang;XU Jian;CONG Feiyun;ZHANG Yidong(Institute of mechanical design and automation,Zhejiang University,Hangzhou 310027,China;Institute of Thermal Science and Power Engineering,Zhejiang University,Hangzhou 310027,China)
出处
《机械设计与研究》
CSCD
北大核心
2018年第4期18-22,共5页
Machine Design And Research
基金
国家自然科学基金(51305392)
浙江省自然科学基金(LZ15E050001)资助项目
流体传动与控制国家重点实验室青年基金资助项目(SKLo FP_QN_1501)
关键词
奇异值分解
K-SVD
自适应学习字典
特征提取
滚动轴承
singular value decomposition
K - SVD
adaptive learning dictionary
feature extraction
rolling bearing
