基于改进层次遗传算法优化的神经网络信道均衡器  

OPTIMISED NEURAL NETWORK CHANNEL EQUALIZER BASED ON IMPROVED HIERARCHICAL GENETIC ALGORITHM

作　　者：薛富强[1] 葛临东[1] 王彬[1] 

机构地区：[1]解放军信息工程大学,河南郑州450002

出　　处：《计算机应用与软件》2010年第5期75-77,88,共4页Computer Applications and Software

基　　金：国防重点预研项目(6130320);河南省基础与前沿研究基金(082300413205)

摘　　要：层次遗传算法的性能与评价个体进化优劣的适应度函数密切相关,对适应度函数的参数进行了研究,提出了依据信道零点和单位圆之间距离设定参数值的新方法,降低了函数复杂度,减少了人为因素影响,提高了算法确定RBF神经网络均衡器结构的效率及性能。仿真结果验证了算法的有效性和稳定性。Since the performance of hierarchical genetic algorithm is closely related to the fitness function used for evaluating the evolution process of individuals,in this paper we studied the parameters of fitness function,and proposed a new algorithm to set their values according to the distance between the zero point of channel and the unit circle of transition function.The new algorithm lowers function complexity and reduces the impact of artificial factors,thus improves the efficiency and performance of the algorithm in determining the structure of RBF neural network equalizer.Simulation results demonstrate the validity and stability of the new algorithm.

关 键 词：层次遗传算法 RBF神经网络 均衡器 适应度函数 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]