基于EMD和STFT柴油机缸盖振动信号时频分析

摘要

A time-frequency analysis method based on empirical mode decomposition (EMD) and short time Fourier transform (STFT) was put forward to research the vibration signal from cylinder head of a diesel engine. The vibration signal was decomposed with self adapting by EMD, and a series of IMF was obtained. Gauss window and Hamming window were chosen for STFT according to different IMF. The time-frequency aggregation index was used as an object function to calculate the best time-frequency distribution of each IMF. Then, these distributions were added altogether to get the best time-frequency distribution of the origin signal. In comparison with other methods, this method can solve the problem of window selection, increase the time-frequency resolution of STFT effectively, and describe time-frequency distribution of vibration signal from the cylinder head exactly.%　　以柴油机缸盖振动信号为研究对象，提出经验模式分解（EMD）和短时傅里叶变换（STFT）相结合的时频分析法。该方法首先利用EMD对缸盖振动信号进行自适应分解，得到一系列本征模态分量（IMF）。根据各分量的特点有针对性地选择高斯窗和汉明窗分别进行STFT，并以时频聚集性指标为目标函数计算各分量的最佳时频分布，最后叠加得到原始信号的最佳时频分布。经对比分析，该方法解决了窗函数及窗宽的选择问题，有效提高STFT的时频分辨率，准确描述柴油机缸盖振动信号的时频分布。