基于改进混合遗传的正交小波盲均衡算法

An Orthogonal Wavelet Transform Constant Modulus Blind Equalization Algorithm Based on Modified Hybrid Genetic Algorithm

为克服传统盲均衡算法收敛速度慢、均方误差大、易陷入局部极小值等缺点,在正交小波变换盲均衡算法(WT-CMA)的基础上,提出一种基于改进混合遗传的正交小波盲均衡算法(MHGA-WT-CMA)。该算法采用基于改进的编码方式、种群初始化、选择算子及交叉算子的遗传算法,以均衡器权系数为初始种群,将正交小波盲均衡算法嵌入遗传算法的父代与子代之间,对父代种群进行局部搜索,将得到的精英个体直接复制到子代中。再将其余个体进行二进制编码、轮盘赌选择、POX交叉以及非均匀变异等遗传算法操作,经过解码成实数进入子代中,进行下一次混合遗传优化,满足停止准则后输出最优权向量。这样可以结合二者的长处,使得算法既能较快收敛,又能在全局范围内得到最优权向量。计算机仿真实验表明,该算法具有收敛速度快、均方误差小、能搜索到全局最优解等特点。

To overcome the disadvantages of Constant Modulus Algorithm（CMA）,such as slow convergence speed,large mean square error,and easily immerging in partial minimum,an orthogonal Wavelet Transform based Constant Modulus blind equalization Algorithm based on Modified Hybrid Genetic Algorithm（MHGA-WT-CMA）was proposed.The proposed algorithm uses genetic algorithm based on modified coding method,population initialization,selection and crossover operator.The coefficient vector of the blind equalizer is regarded as the initial population.Orthogonal wavelet blind equalization algorithm is embedded into the genetic algorithm to search elite individual in the father generation population locally.The elite individual is copied into the offspring directly.The other individuals are optimized by modified genetic algorithm with binary coding method,roulette wheel selection method,POX crossover and non-uniform mutation.They are sent into offspring after decoding into real.The hybrid genetic algorithm outputs the optimal weight vector when it satisfies stopping criterion.The proposed algorithm combines advantages of both orthogonal wavelet transform based constant modulus blind equalization algorithm and modified genetic algorithm.Computer simulation shows that the proposed algorithm has fast convergence rate,small mean square error,and global optimal solution.