适用于时变系统的在线盲源分离算法研究

针对传统EASI算法收敛速率与稳态误差之间的矛盾，提出了一种基于估计函数期望的步长自适应算法（New Adaptive EASI），为了使这种算法能够更好地解决时变系统中不同条件下的盲源分离问题，提高信号的分离精度，建立了一种混合矩阵变化的在线检测机制，并将这种在线检测机制加入步长自适应算法中，对算法进行了改进。仿真实验表明，这种改进的步长自适应算法能够提高盲源分离初始阶段或是信道变化后分离初始阶段的信号恢复质量，解决源信号为非零均值信号时的盲源分离问题，并且能够准确地在线估计源信号的个数，实现信源数变化条件下的盲源分离。

Research on on-line blind source separation algorithm for time-vary-ing system

Aiming at the contradiction between the convergence rate and the steady-state error of the traditional EASI algorithm, this paper presents a New Adaptive EASI algorithm based on the expectation of estimate function. In order to make the new algorithm be able to solve the BSS problems under different conditions of time-varying system better and improve the accuracy of separated signals, it sets up an on-line detection mechanism of the changes in mixing matrix and adds it to the New Adaptive EASI algorithm to modify the algorithm. Simulation results show that the modified New Adaptive EASI algorithm can improve the quality of separated signals in the initial stage of separation or just after the channel condition changes, do the on-line separation of nonzero mean mixed signals and accurately estimate the number of source signals online to solve the BSS problem under the condition that the number of the source signals varies during the separating process.