符号熵与证据理论集成的转子故障识别研究  

RESEARCH ON ROTOR FAULT IDENTIFICATION BASED ON THE INTEGRATION OF SYMBOLIC ENTROPY AND EVIDENCE THEORY

作　　者：山崧[1] 陈泓铭 刘靓璇 邓帅卿 苏宇 SHAN Song;CHEN HongMing;LIU JingXuan;DENG ShuaiQing;SU Yu(Tianjin Branch Company,SINOPEC,Tianjin 300271,China;School of Mechanical Engineering,Xi′an Jiaotong University,Xi′an 710049,China)

机构地区：[1]中国石油化工股份有限公司天津分公司,天津300271 [2]西安交通大学机械工程学院,西安710049

出　　处：《机械强度》2024年第3期540-549,共10页Journal of Mechanical Strength

摘　　要：针对转子振动信号的非线性、非平稳性造成的故障状态难以准确识别问题,提出了一种符号熵与证据理论集成的转子故障识别方法。首先对转子系统的监测数据进行去均值处理和差分运算得到波动序列,继而通过二进制处理得到二进制序列,再通过编码得到符号序列而计算符号熵。通过一个符号熵均值相对增长率的指标来确定符号熵计算中的编码长度。通过实验分析,以转子四个通道的符号熵作为识别向量,再与历史数据得到的识别标准向量做对比,得到各状态的识别概率,最后通过D-S证据理论将其中两次测试的结果进行融合得到最终的判别结果。通过与传统的基于符号熵的故障识别方法的对比,能准确识别转子系统的常见状态,为振动信号处于非线性、非平稳条件下的转子故障状态识别问题提供了一个解决方案。A rotor fault identification method integrating symbolic entropy and evidence theory was proposed for the problem of the difficulty to accurately identify the fault state caused by the nonlinearity and nonstationarity of the rotor vibration signal.Firstly,the fluctuation sequence was obtained by processing the recorded data of the rotor system by de-averaging and difference.Secondly the binary sequence was obtained by binary processing.Thirdly the symbol sequence was obtained by encoding to calculate the symbol entropy.In this paper,the code length in the calculation of symbol entropy was determined by an indicator of the relative growth rate of the mean value of symbol entropy.Through experimental analysis,the symbol entropy of the four channels of the rotor was used as the identification vector,and then compared with the identification standard vector obtained from the historical data to obtain the recognition probability of each state,and finally the results of two tests were fused through D⁃S evidence theory to obtain the final discriminatory result.Compared with the traditional fault identification method based on symbolic entropy,the proposed method can accurately identify the common states of the rotor system,providing a solution to the rotor fault state identification problem when the vibration signals are in nonlinear and nonstationary conditions.

关 键 词：转子 符号熵 证据理论 故障识别 相对增长率 

分 类 号：TH133[机械工程—机械制造及自动化] TH17