一种强鲁棒性的盲分离混合矩阵估计方法

A Robust Method for Mixing Matrix Estimation in Blind Source Separation

针对传统盲分离混合矩阵估计鲁棒性差、易受初始值影响、精度不高等问题,该文将人工蜂群算法(ABC)用到盲分离中,结合稀疏信号混合矩阵估计的特点,提出一种基于不同搜索策略和编码方式的两阶段蜂群算法的混合矩阵估计方法,通过新的蜜蜂搜索行为和子蜂群之间的协同作业,明显加快了算法的收敛速度,提高了混合矩阵的估计精度。仿真实验表明,该方法在源个数较多、弱稀疏、低信噪比的情况下仍然可以很好地估计混合矩阵。相比已有方法,该方法不仅具有很强的鲁棒性和很高的估计精度,而且不需要太大的计算量。

To solve the problems of traditional methods for mixing matrix estimation in blind source separation such as poor robustness, the defect that separation performance is vulnerable to the initial value and low estimation accuracy, Artificial Bee Colony （ABC） algorithm is applied to blind source separation. Combining with the characteristics of mixing matrix estimation for sparse signals separation, a two-stage bee colony algorithm based on different searching strategy and encoding mode is proposed to estimate mixing matrix, which can accelerate the convergence rate and enhance estimation precision through bees＇ new searching behavior and collaboration between the bee colonies. The simulation results show that the proposed method can perform very well even in the case of large-scale, weak sparse and low SNR. The proposed method not only has the characteristics of strong robustness and high estimation accuracy compared with existing methods, but also need not a large amount of calculation.