一种求解优化问题的新型小波混沌神经网络  

Novel wavelet chaotic neural network for solving optimization problems

作　　者：孙明[1] 赵琳[1] 徐耀群[2] 丁继成[1] 

机构地区：[1]哈尔滨工程大学自动化学院,哈尔滨150001 [2]哈尔滨商业大学自动化学院,哈尔滨150028

出　　处：《计算机工程与应用》2010年第11期42-45,共4页Computer Engineering and Applications

基　　金：国家自然科学基金No.60474046;国家教育部新世纪人才支持计划No.NCET-04-0339~~

摘　　要：基于神经元的自反馈项可产生混沌的现象,将Gauss小波函数作为混沌神经元的自反馈项。分析了Gauss小波的尺度和平移参数对神经元动力学的影响,提出了自反馈连接权和Gauss小波尺度双重模拟退火的混沌神经元。构建了混沌神经网络模型,分析了由Gauss小波函数产生的附加能量函数对网络优化能力的影响。优化问题的仿真结果表明,该网络能够以较快的速度找到优化问题的全局最优解。Based on the phenomena that the neuron＇s self-feedback can generate chaos,Gauss wavelet function is used as the self-feedback of chaotic neuron in this paper.The effect of the scale and shift parameters in Gauss wavelet on the neuron dynamics is analyzed and the chaotic neuron with double simulated annealing mechanisms realized respectively by the self-feedback connection weight and the Gauss wavelet scale is proposed.The chaotic neural network is constructed and the effect of the additional energy function generated by Gauss wavelet function on the optimization ability of the network is studied.The simulation results on optimization problems show that the proposed network can find the global optimum of the optimization problems with a faster speed.

关 键 词：Gauss小波 混沌神经网络 优化问题 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]