基于蚁群算法和互信息进行非线性盲源信号分离

Application of Ant Colony Algorithm and Mutual Information in Nonlinear Blind Source Separation

针对在非线性混叠盲源分离中代价函数往往具有许多局部最优解,而求解其全局最优解非常困难的问题,这里提出一种基于蚁群算法进行非线性盲源信号分离的方法.该方法用高阶奇数多项式拟合非线性混合函数,以分离信号的互信息作为代价函数.并对非线性混合信号进行了仿真研究.由此方法得到的分离信号和源信号的相似系数都在98%以上,仿真实验结果表明该方法能够对非线性盲源信号进行很好地分离.

In nonlinear blind source separation,usually it is very difficulty to find the global optimal values of cost functions due to the existence of many local optimal values.In order to overcome the disadvantages mentioned above,the method of nonlinear blind source separation based on ant colony algorithm is proposed in this paper.The high-order odd polynomial function is used to fit the nonlinear transfer function in this method;the mutual information of separation signals is used as cost function of ant colony algorithm.In simulation experiment,nonlinear mixed signals are successfully separated by this method.The similar coefficients of separation signals obtained by this algorithm and source signals are higher than 98%.The simulation experiment results have shown that this method can well solve the problem of nonlinear blind source separation.