基于EEMD 和SVM 的滚动轴承退化状态识别 - 计算机集成制造系统.PDF

基于EEMD 和SVM 的滚动轴承退化状态识别 + 魏永合 , 王明华 （沈阳理工大学机械工程学院，辽宁 沈阳 110159） 摘要：为准确识别滚动轴承退化状态，提出了一种集合经验模态分解 （EEMD ）和支持向量机 （SVM）相结合的方法进行滚动轴承的退化状态识别方法。采用EEMD 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（GA ）和SVM 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建立退化状态识别模型。最后从不同 退化状态的测试数据中提取出经过 GA 优化删选后的特征向量，将其输入参数经过 GA 优化过的 SVM 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。 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 中图分类号：TH17 文献标识码：A Degradation state recognition of rolling bearing based on EEMD and SVM WEI Yong-he+，WANG Ming-hua (School of Mechanical Engineering ,Shenyang Ligong University, Shenyang 110159,China) Abstract: In order to accurately recognize the degradation state of rolling bearing, a hybrid method combining Ensemble Empirical Mode Decomposition (EEMD) method and a Support Vector Machine (SVM) was proposed, and the model for degradation state recognition of rolling bearing was constructed. Firstly the original signal was decomposed into many components via EEMD adaptively，and adaptive reconstruction was performed by using the correlation coefficient method to eliminate noise and fault diagnosis. The feature vectors of degradation state were extracted through the combination of Genetic Algorithm(GA) and SVM. Then the degradation state probability distribution and historical remnant life of rolling bearing are calculated to determine the optimal number of degradation state, which is employed to construct the SVM model for degradation state recognition. The analytical results for full lifetime datasets of a certain bearing demonstrate the validity of the method. Key words:ensemble empirical mode decomposition; genetic algorithm; support vector mach