基于特征熵和优化SVM的轴承故障诊断方法  

Bearing fault diagnosis method based on characteristic entropy and optimized support vector machine

作　　者：吴国洋[1] 

机构地区：[1]攀枝花学院材料工程学院,四川攀枝花617000

出　　处：《兰州理工大学学报》2013年第4期32-36,共5页Journal of Lanzhou University of Technology

基　　金：国家科技支撑计划(2006BAA01A11)

摘　　要：为了对轴承的故障进行有效的识别,提出基于特征熵和优化支持向量机的轴承故障识别新方法.利用EMD分解信号提取分解信号的能量熵,由于这些熵值之间冗余信息较为严重,因此选用主成分分析对这些熵信息进行约简,提取最有效的特征信息,作为支持向量机模型的输入.通过粒子群优化选取最优决策树构造最佳的支持向量机分类模型进行状态的识别和判定,提高了分类的精确度.通过一个滚动轴承的实例说明方法的有效性和准确性.In order to carry out effective identification of bearing running fault, a new method of bearing fault identification was proposed based on characteristic entropy and optimized support vector machine. The empirical mode decomposition method was used to extract the signal energy entropy. Because the re- dundant information among these entropy values still exist and very serious, the principal component anal- ysis was selected to conduct the reduction of these entropy information and extract the most effective char- acteristic information as the input to the support vector machine model. By means of optimization of parti- cle swarm, the optimal decision making tree was chosen, the classification model of optimal support vector machine was constructed, and state identification and judgement were performed, so that the classification accuracy was improved. A rolling bearing example was given to illustrate the effectness and accuracy of the method.

关 键 词：轴承 故障诊断 主成分 粒子群 支持向量机 

分 类 号：TH165.3[机械工程—机械制造及自动化]