摘要
分析了带有周期性突变成分信号的频域特征,提出了识别周期性突变的倒变异高阶谱方法,该方法首先对信号的FFT所得数据进行调整,并根据周期性突变的特点设计了新的高阶谱频率函数,由频率函数可得到变异高阶谱,进而对变异高阶谱的某些频率取纵向切片计算倒谱,得到倒变异高阶谱.基于倒变异高阶谱的周期性突变识别方法可充分利用FFT所得的正负频率域的所有数据,扩充了高阶谱的生成空间,提高了对周期性突变的识别效果.试验表明所提方法可准确识别周期性突变和突变的周期,具有一定的实际应用价值.
The frequency domain characteristics of signals with periodic breaks are analyzed. A novel identification method with variational higher order cepstrum is presented. In this method, the result data of Fourier transform is adjusted firstly, then the variational higher order spectrum is calculated according to the new frequency function designed for improving the analysis effect. To estimate the period of breaks, the cepstrum, namely the variational higher order eepstrum of some frequency slice in above spectrum should be calculated. The proposed method can utilize whole data of Fourier transform to enlarge the space for calculating higher order spectrum, and improve the effect of identifying the periodic breaks. The results of experiments show that this method can identify the periodic breaks and the period of breaks accurately.
出处
《江苏大学学报（自然科学版）》
EI
CAS
北大核心
2008年第2期147-150,共4页
Journal of Jiangsu University：Natural Science Edition
基金
国家自然科学基金资助项目(50775029)
辽宁省教育厅科技攻关项目(2004D089)
关键词
机械工程
周期性突变
高阶谱
故障诊断
特征提取
倒谱
信号识别
mechanical engineering
periodic breaks
higher order spectrum
faults diagnosis
signature extracting
cepstrum
signal identification
