单通道盲源分离算法及其在工程机械振源分析中的应用.pdfVIP

第 52 卷第 10 期 2 0 1 6 年 5 月 机 械 工 程 学 报 JOURNAL OF MECHANICAL ENGINEERING Vol.52 No.10 May 2 0 1 6 DOI：10.3901/JME.2016.10.001 单通道盲源分离算法及其在工程机械 振源分析中的应用* 于 刚 周以齐 (山东大学高效洁净机械制造教育部重点实验室 济南 250061) 摘要：研究一种新的单通道盲源分离方法，解决了传统盲源分离方法因传感器数量不足而无法有效分离源信号的问题，同时 源信号幅值也得到了准确的恢复。首先利用集合经验模态分解方法将一维测量信号分解为具有不同尺度特征的本征模态函 数，而后与原测量信号组成多个二维矩阵，通过稀疏分量算法得到各源信号的真实估计。利用仿真信号，与已有方法进行对 比，验证了提出方法的有效性。将提出的方法应用在挖掘机动力源附近的振动分析中，成功分离出了多个振源信号。利用时 频分析对源信号分别进行特征识别，并将分离结果用于振源的贡献度与声信号传递特性分析。得到各振源对于测试位置的贡 献度排序，以及对于挖掘机噪声信号的传递规律，为挖掘机的减振降噪措施提供了可靠的依据。 关键词：单通道盲源分离；挖掘机；振源贡献量分析；传递特性分析 中图分类号：TH17 Single-channel Blind Source Separation and Its Application in Analyzing Vibration of Engineering Machinery YU Gang ZHOU Yiqi (Key Laboratory of High-efficiency and Clean Mechanical Manufacture Shandong University, Ministry of Education, Jinan 250061) Abstract：A novel blind source separation method is proposed, which can separate sources with single channel measurement signal. And the amplitude of sources can be recovered accurately. First of all, one-dimensional measurement signal is decomposed into several intrinsic mode functions (IMF) by ensemble empirical mode decomposition, and combined with corresponding IMF into multiple two-dimensional matrixes. Then the sources can be separated by sparse component analysis method effectively. The effectiveness of proposed method is verified by simulated data. In practical application, the proposed method is utilized for separating sources by one-dimensional measurement signal recorded from an excavator. Then based on the separated sources, the contribution characteristics and transfer behaviors are analyzed. The analysis results can be utilized to guide the reduction of excavator vibration and noise. Key words：single-channel source separation；excavator；contribution estimation；transfer analysis 0 前言* 振动直接关系到机械系统的性能与寿命，并对 操作人员的身体健康和操作安全产生影响 [1] 。工程 机械中振源较多，典型振源包括内燃机、液压泵、 进气、排气及冷却系统等，各振源到驾驶室、工作 机构等目标位置的传递路径复杂，导致测量信号通 常为多个激励的叠加。有必要针对各