基于改进频域压缩感知的轴承复合故障欠定盲提取

Underdetermined blind extraction for bearings complex failures based on improved frequency domain compressive sensing

针对旋转机械复合故障频域盲提取算法的不足,为提高欠定盲提取分离结果精度,提出基于多尺寸多结构元素的闭-开组合形态滤波、遗传模拟退火聚类及频域压缩感知重构算法相结合的(C-OACMF-GASA-CS)故障特征欠定盲提取方法。利用形态滤波滤除背景噪声及提取冲击信号;利用遗传模拟退火算法估计混合矩阵;用估计矩阵重构传感矩阵,并用正交匹配追踪基频域压缩感知重构分离信号。实验仿真及双通道滚动轴承故障加速度振动信号分析结果表明,该方法能有效分离提取滚动轴承故障特征。

Considering the disadvantges of frequency-domain blind extraction algorithm for complex rotating machinery faults,and improving the separation accuracy of underdetermined blind extraction,a method integrating the multi-scale multi-structure close-open average combination morphological filtering （C-OACMF ）,genetic simulated annealing clustering （GASA）and frequency domain compressive sensing（CS）algorithm was proposed to deal with the underdetermined blind fault feature extraction.The C-OACMF was used to filter out the background noise and extract the shock signals;the GASA of fuzzy C-average clustering algorithm was used to estimate the mixed matrix;the sensor matrix was reconstructed with the help of the estimated matrix and the orthogonal matching pursuit of frequency domain CS algorithm was used to estimate the source signals.The results of computer simulation and real rolling bearing signals analysis show that the proposed method is quite effective.