摘要
为解决旋转机械智能诊断中缺少量化特征值的问题,提出了一种基于小波分析和矩不变量的量化特征提取新方法。在对实验转子振动信号进行小波分析的基础上,采用了灰度图来对连续小波分解系数进行表达,并提取出灰度图的7个矩不变量作为描述振动信号状态特征的量化特征值系列。分析表明,该方法所提取的转子信号量化特征值系列,能够有效地区分出几种典型故障间的差别。
In order to solve the problems for lack of quantized feature value in the intelligent diagnosis of rotary machinery, a new method was proposed to extract the quantized features based on wavelet analysis and moment invariants. Based on wavelet analysis of the vibration signals of the experimental rotors, a grey image was used to express the continuous wavelet decomposition coefficients, and seven moment invariants of the grey image were extracted as the quantized feature value series of describing the condition features of the vibration signals. The analysis shows that the quantized feature value series of the rotor signals extracted using the new method can efficiently identify the differences among the typical faults.
出处
《化工机械》
CAS
2007年第6期309-312,共4页
Chemical Engineering & Machinery
基金
甘肃省科技攻关计划资助项目(2GS064-A52-035-02)
关键词
转子
故障诊断
特征提取
小波分析
矩不变量
Rotor, FaUlt Diagnosis, Feature Extraction, Wavelet Analysis, Moment Invariant
