基于线性混合盲分离的谐波检测和消除方法

Harmonic Detection and Removal by Linear Blind Separation

盲分离的信号一般具有尺度和次序不确定性,因此应用于谐波检测时得到的谐波参数可能是不准确的。针对该问题,提出了基于线性混合盲分离的谐波检测和消除方法。该方法在传统基于独立分量分析的谐波检测和消除方法的基础上,采用矩阵方法分析了线性混合盲分离模型中分离信号与源信号的对应关系,得到了一种消除分离信号尺度和次序不确定性的方法,使传统的线性盲分离模型应用于谐波检测时得到的谐波参数更加准确,仿真实例证明了该方法在谐波幅度检测和谐波消除方面的正确性和有效性。

The scale and the permutation of blind source separation are ambiguous, so the harmonic detection based on linear blind separation is inaccurate. Aimed at the problem, the thesis put forward an algorithm of har- monic detection and removal by linear blind separation. On the foundation of traditional ICA algorithm for har- monic estimation and removal by linear blind separation, the corresponding relation between source signal and separated signal was discussed by utilizing the matrix analysis, and an algorithm was presented to eliminate am- biguity of the scale and permutation. By using the algorithm, harmonic parameters detection with linear blind source separation model was more accurate. Simulations results demonstrated that the proposed algorithm was correct and effective in magnitude detection and harmonic removal.