一种盲源提取方法及其在滚动轴承故障特征提取中的应用

发布时间：2018-04-27 03:26

本文选题：故障诊断 + 盲源提取　；参考：《振动工程学报》2014年05期

【摘要】：利用滚动轴承发生故障时其故障信号往往呈现出一定周期性的特点,首先计算出故障信号的理论基本周期τ,将τ作为待提取源信号的基本周期用所述方法的相关步骤计算出大致目标源信号及权重分离矩阵^W。然后将^W作为初始权重分离矩阵,将基于高阶统计量的固定点算法用于原始观测信号提取出更为精确的目标故障信号。通过仿真信号和实验信号验证了所述方法相对于约束独立成分分析(Constrained independent component analysis,CICA)方法具有以下优点:不需要精确估计目标源信号的周期及不需要构建精确的参考信号。此外,通过仿真还验证了所述方法相对于其他较新的盲源提取方法具较高的提取精度等优点。
[Abstract]:When the rolling bearing is malfunction, the fault signal often presents a certain periodic characteristic. First, the theoretical basic period of the fault signal is calculated. As the basic period of the source signal to be extracted, the target source signal and weight separation matrix ^W. are calculated by the relevant steps of the method described, and then the ^W is used as the initial weight division. The fixed point algorithm based on high order statistics is used to extract more accurate target fault signals from the original observation signal. The following advantages are verified by the simulation signal and the experimental signal. The Constrained independent component analysis (CICA) method has the following advantages: no need to be accurately estimated. The cycle of the target source signal and the construction of an accurate reference signal are not needed. In addition, the advantages of the proposed method are proved to be higher than other new blind source extraction methods.

【作者单位】：上海交通大学机械系统与振动国家重点实验室;
【基金】：国家自然科学基金资助项目(51035007,51105243)
【分类号】：TH165.3;TN911.7

【共引文献】