Abstract:

To improve the convergence speed of independent component analysis with reference, an improved fast convergence reference independent component analysis method is proposed. The observed signals are firstly processed by pre-whitening method to reduce the complexity of algorithm. Secondly, Newton iteration method with faster convergence speed and simpler structure are used to optimize the algorithm and improved its convergence speed. The theoretical analysis shows that the computational complexity of the improved algorithm is almost the same as that of the existing algorithm in each iteration. The simulation results show that compared with the existing methods, the improved algorithm has faster convergence speed with the same error.

Key words: independent component analysis with reference, Newton iteration method, convergence speed, complexity of algorithm

摘要：

为提高参考独立分量分析收敛速度，提出了改进的快速收敛参考独立分量分析方法。该方法首先采用预白化方法对观测到的信号进行处理，降低后续处理算法的复杂度，然后采用收敛速度更快、结构更简单的牛顿迭代方法对其进行优化，提高其收敛速度。理论分析表明，改进后的算法相对现有算法每次迭代的计算量基本相同；实验仿真结果表明，相对现有方法，改进后的算法具有较快的收敛速度，且误差保持不变。

关键词: 参考独立分量分析, 牛顿迭代方法, 收敛速度, 算法复杂度