基于EEMD-Fast ICA-STFT的车用起动电机噪声源识别

摘要

提出基于集成经验模态分解(EEMD)、快速独立分量分析(Fast ICA)和短时傅里叶变换(STFT)的噪声源识别方法,对起动电机噪声信号进行声源识别研究.首先采用集成经验模态分解法将单一通道的电机噪声信号分解为一系列本征模态分量,随后用Fast ICA算法提取独立成分,最后利用短时傅里叶变换良好的时频分析特性,对Fast ICA分离结果进行时频分析,结合时频分析结果和电机噪声的先验知识,确定了各独立分量与电机不同噪声源的对应关系.%A noise source identification method based on the ensemble empirical mode decomposition (EEMD), fast independent component analysis (Fast ICA) and short time Fourier transform (STFT) algorithms is proposed to study the noise source identification of vehicle's starting motors. First of all, the EEMD algorithm is used to decompose the single channel noise of the starting motors into several intrinsic mode functions. Then, the Fast ICA algorithm is used to extract the independent components. Finally, using the better time-frequency characteristics of STFT algorithm, the time-frequency characteristics of the Fast ICA results are analyzed. Combining the results with the prior knowledge of the motor noise, the relationship between the independent components and the different noise sources of the motors is determined.