ISVD降噪和重排谱图在烟机信号HHT时频谱分析中的应用

Application of ISVD de-noising and reassigned spectrogram in HHT time-frequency spectrum analysis of flue gas turbine signal

针对基于希尔波特黄变换(HHT)的时频谱存在受噪声影响大、高频部分时频分辨率低的缺点,提出将迭代奇异值分解(ISVD)降噪和重排谱图应用于HHT时频谱分析中。首先对噪声对HHT时频谱的影响进行了分析,含噪实测信号经过经验模式分解(EMD)得到的基本模式分量(IMF)中含有很强的噪声成分,从中提取的瞬时参数不具有物理意义,构成的HHT时频谱混乱而不准确;然后利用ISVD降噪对实测信号进行了降噪处理,有效地去除了噪声,降噪后信号EMD分解得到的IMF分量其Renyi信息接近于0,是近似平稳的单分量信号,从中可以提取到准确的瞬时参数,构成特征清晰、分布合理的HHT时频谱;最后对HHT时频谱中高频部分所对应的IMF分量进行了重排谱图分析,实现了其时频特征的准确定位。结果表明,该方法可以有效地提高烟气轮机信号HHT时频谱分析的准确率,为故障诊断提供可靠的特征依据。

In order to improve the quality of time-frequency spectrum based on Hilbert-Huang transform （ HHT）, the iterative singular value decomposition （ISVD） de-noising and reassigned spectrogram are applied in this paper. Firstly, the influence of noise is discussed. Using empirical mode decomposition （ EMD）, the intrinsic mode function （IMF） obtained from the noise-contained signal contains a lot of noise. So the instantaneous parameters obtained from IMF are not correct and the HHT time-frequency spectrum has no physical meaning. Then the iterative singular value decomposition （ISVD） de-noising is applied. Through ISVD de-noising, noise is reduced effectively and the Renyi information of the IMF is near zero. As the IMF becomes single component signal, the obtained instantaneous parameters are accurate and the HHT time-frequency spectrum is reasonable. Finally, because the HHT time-frequency spectrum has low time-frequency resolution in high-frequency domain, the IMF relevant to this domain is analyzed using reassigned spectrogram. The time-frequency feature is extracted accurately. Results prove that the method proposed in this paper is effective and can improve the accurate rate of fault diagnosis of flue gas turbine.