基于粒子群优化的后向非线性独立分量分析算法

Analyzing post-nonlinear independent component by particle swarm optimization algorithms

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 particle swarm optimization, a novel method is proposed to minimize the mutual information. The nonlinear transfer function was simulated by the P-th order polynomial function. After analyzing the shortcoming of the particle swarm optimization algorithm, the velocity updating formula by adding the disturbance term was modified. Through improving position vector and velocity vector, we get the global optimization solution and then separated the mixed signals. The simulation results showed that the independent component analysis based on particle swarm optimization had more remarkable performance.