期刊
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
卷 70, 期 -, 页码 12-12出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TIM.2020.3045841
关键词
Transforms; Time-frequency analysis; Fault diagnosis; Vibrations; Fourier transforms; Analytical models; Tools; Fault diagnosis; instantaneous frequency (IF); nonstationary signal; synchroextracting transform; synchrosqueezing transform (SST); time-frequency (TF) analysis
The article introduces two methods, synchrosqueezing transform (SST) and second-order synchroextracting transform (SET2), for improving the TF representation of nonstationary signals. While SST uses the GMLC signal model to accurately represent time-varying nonstationary signals, SET2 focuses on characterizing TF distribution using energy at the IF to reduce the impact of noise on the real signal.
Synchrosqueezing transform (SST) is a currently proposed novel postprocessing time-frequency (TF) analysis tool. It has been widely shown that SST is able to improve TF representation. However, so far, how to improve the TF resolution while ensuring the accuracy of signal reconstruction is still an open question, particularly for the vibration signal with time-varying instantaneous frequency (IF) characteristics, due to the fact that the vibration signals of mechanical equipment usually contain many types of noise generated by harsh operating conditions, and the SST will mix these noise into the real signal. Our first contribution is using the Gaussian modulated linear chirp (GMLC) signal model to represent the general nonstationary signals. The GMLC signal model can more accurately represent the time-varying nonstationary signal, compared with the SST signal model composed of linear phase function and constant amplitude. Our second contribution in this work is proposing a method to improve the TF resolution and reconstruction accuracy for nonstationary signals with time-varying IF, which we coined the second-order synchroextracting transform (SET2). In SET2, we apply the GMLC to deduce the nonstationary signal model and then only use the energy at the IF to characterize the TF distribution, which improves the TF while reducing the impact of noise on the real signal.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据