基于EEMD云模型与SVM的汽轮机转子故障诊断方法  

A rotor fault diagnosis method based on EEMD cloud model and SVM

作　　者：田松峰[1] 胥佳瑞 王美俊[1] 韩强[1] 

机构地区：[1]华北电力大学电站设备状态监测与控制教育部重点实验室,河北保定071003

出　　处：《热力发电》2017年第4期111-114,共4页Thermal Power Generation

基　　金：中央高校基本科研业务费专项资金资助项目(12QN39);河北省自然科学基金资助项目(E2012502016)~~

摘　　要：提出了一种基于集合经验模态分解(EEMD)、云模型与支持向量机(SVM)相结合的汽轮机转子多故障诊断方法。该方法首先采用EEMD将振动信号分解成若干个IMF分量,利用相关系数法对IMF分量进行筛选,然后对筛选后的IMF分量进行逆向云发生器计算,得到云模型的数字特征并构建为特征向量,将其应用到有向无环图SVM中进行转子多故障状态识别,并与传统的EEMD能量法进行对比。结果表明,该方法能够准确地完成转子多故障诊断,具有更高的识别率。A multi-fault diagnosis method for steam rotors based on ensemble empirical mode decomposition （EEMD）, cloud model and support vector machine （SVM） was proposed. This method firstly uses the EEMD to decompose the vibration signals into several IMF components and the correlation coefficient method to screen the IMF components. Then, it applies the backward cloud generator algorithm to compute the screened IMF components and obtains digital characteristics of the cloud model to build the feature vector. Finally, the feature vector is applied to the directed acyclic graph SVM to perform multi-fault state recognition for the rotors, and the result is compared with that of the conventional EEMD energy method. The experimental results show that, this method can accurately complete multi-fault diagnosis for the rotors and has more accurate recognition rate.

关 键 词：转子 故障诊断 集合经验模态分解 云模型 支持向量机 特征向量 

分 类 号：TK268.1[动力工程及工程热物理—动力机械及工程]