基于核典型相关分析的非线性相关源盲分离方法研究

Blind separation for nonlinear mixture of correlated sources based on kernel canonical correlation analysis

针对传统的相关源盲分离方法的不足,提出了一种基于核典型相关分析的非线性相关源盲分离方法。该方法是利用了核方法来处理数据之间的非线性问题,同时还利用信号源之间的相关性来进行分离。提出的方法与传统的相关源盲分离方法进行对比分析。仿真结果表明,提出的方法明显优于传统的相关源盲分离方法,并从分离性能指标上得到了充分的反映。最后,将该方法应用到转子不对中和碰摩故障的盲分离中,实验结果进一步验证了该方法的有效性。

Based on the deficiency in the traditional blind separation method of statistically correlated sources,a new blind separation method of nonlinear mixture from correlated sources is proposed. In the proposed method,the nonlinear problem between the data is processed using the kernel method,and the correlated sources is effectively separated using the correlation of source signals. The proposed method is compared with traditional blind separation method of statistically correlated sources. The simulation results show that the proposed method is obviously superior to the traditional blind separation method of correlated sources,and the effectiveness of separation has been fully reflected with the performance index of separation. Finally, the proposed method has been applied in blind separation of the misalignment of rotary and rotor rub-impact,the experiment results further verify the effectiveness of the proposed method.