论文部分内容阅读
在分析后向非线性混合独立分量分析算法的基础上,提出了一种基于粒子群优化的独立分量分析算法.该算法以互信息量最小化为目标函数,用高阶奇数多项式拟合非线性分离函数,针对现有粒子群算法的不足,引入带有扰动项改进速度更新公式,通过对粒子群位置矢量和速度矢量的更新,得到全局最优值,从而得到分离矩阵和分离多项式参数.仿真结果表明所提算法是一种非常有效的盲源分离算法.
Based on the analysis of back-mixing non-linear mixed independent component analysis algorithm, an independent component analysis algorithm based on particle swarm optimization is proposed, which minimizes the mutual information as the objective function and high-order odd polynomials to fit nonlinearity Aiming at the insufficiency of the existing PSO, the improved velocity update formula with perturbation term is introduced, and the global optimal value is obtained by updating the particle swarm position vector and velocity vector to obtain the separation matrix and the separation polynomial parameters. The results show that the proposed algorithm is a very effective blind source separation algorithm.