摘要
为准确识别滚动轴承退化状态,提出一种集合经验模态分解和支持向量机相结合进行滚动轴承的退化状态识别方法。采用集合经验模态分解对原始信号进行分解、降噪、信号重构和故障类型诊断,通过遗传算法和支持向量机优化提取状态识别特征,利用滚动轴承退化状态概率分布以及历史剩余寿命来确定其最优退化状态数目,以建立退化状态识别模型。从不同退化状态的测试数据中提取出经过遗传算法优化删选后的特征向量,将其输入用遗传算法进行参数优化的支持向量机中进行退化状态的识别分类。实验结果表明,该方法可以实现滚动轴承退化状态的准确识别。
To recognize the degradation state of rolling bearing accurately, a hybrid method by integrating Ensemble Empirical Mode Decomposition （EEMD） and Support Vector Machine （SVM） was proposed, and the model for degradation state recognition of rolling bearing was constructed. The original signal was decomposed into many com- ponents with EEMD adaptively,and adaptive reconstruction was performed by using the correlation coefficient meth- od to eliminate noise and fault diagnosis. The feature vectors of degradation state were extracted through the combi- nation of Genetic Algorithm （GA） and SVM. The degradation state probability distribution and historical remnant life of rolling bearing were calculated to determine the optimal number of degradation state, which was employed to construct SVM model for degradation state recognition. The analytical results for full lifetime datasets of a certain bearing demonstrated the validity of the method.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2015年第9期2475-2483,共9页
Computer Integrated Manufacturing Systems
基金
国家863计划资助项目(2012AA041303)~~
关键词
集合经验模态分解
遗传算法
支持向量机
滚动轴承
退化状态
故障诊断
ensemble empirical mode decomposition
genetic algorithms
support vector machine
rolling bearing
degradation state
fault diagnosis
