基于微粒群算法的盲解卷积及其在故障轴承声信号中的应用

PSO-based Blind Deconvolution Algorithm and Its Application in Bearing Fault Detection by Acoustical Analysis

针对复杂机械声场中,盲解卷积算法分离滤波器长度估计困难的问题,提出一种用微粒群算法优化盲解卷积分离滤波器长度的改进方法,解决了估计结果不确定的问题。该方法以分离信号峭度最大值的倒数为微粒群搜索的适应度,利用微粒群算法搜索到最优的分离滤波器长度,获得了可靠稳定的估计结果。计算机仿真和实际环境中故障轴承声信号分离实验证明了该方法的有效性。

In complex mechanical sound field,it is very difficult to set the length of separation filters of blind deconvolution algorithm which generally leads to the separation results uncertain.Here,an improved blind deconvolution based on PSO algorithm was proposed.The reciprocal of maximize kurtosis was used to be as fitness function,and the optimal length of the separation filters can be selected by the PSO algorithm.With the help of the enhancement,good results can be obtain.Results of simulation and acoustical signals of faulty rolling bearings measured in actual environment were presented to demonstrate that the method is availability.