基于VMD和改进的FastICA算法的旋转机械故障信号分离方法研究

针对分离出的旋转机械故障信号的非线性非平稳性问题,本文提出一种对旋转机械故障信号分离的方法.首先针对以往利用EMD方法分解的特征信号存在的模态混叠问题,利用VMD方法完成对旋转机械故障特征信号的分解.其次,选取相应的分解后特征信号构成观测序列,利用FastICA算法对观测序列进行分离得到源信号,最后针对FastICA算...

Study on the Separation Method of Rotating Machinery Fault Signal Based on VMD and improved Fast ICA algorithm

Aiming at the nonlinear and non-stationary problem of the separated rotating machinery fault signals, this paper proposes a method to separate the rotating machinery fault signals. First, aiming at the problem of modal aliasing existing in the characteristic signals decomposed by the EMD method, the VMD method is used to complete the decomposition of the characteristic signals of the rotating machinery. Secondly, select the corresponding decomposed characteristic signals to form the observation sequence, and use the FastICA algorithm to separate the observation sequence to obtain the source signal. Finally, for the shortcomings of the FastICA algorithm such as poor convergence and sensitivity to the initial value, a method based on VMD and improved FastICA is proposed. Algorithm-based method for separating fault signals of rotating machinery. Experiments show that this method improves the convergence speed of FastICA algorithm and improves the quality of fault signal separation of rotating machinery.