一种用于函数优化的小波混沌神经网络  

Wavelet chaotic neural network for function optimization

作　　者：周婷[1] 贾振红[1] 刘秀玲[1] 

机构地区：[1]新疆大学信息科学与工程学院,乌鲁木齐830046

出　　处：《计算机应用》2007年第12期2910-2912,共3页journal of Computer Applications

基　　金：教育部新世纪优秀人才支持计划资助项目(NCET-05-0897);新疆维吾尔自治区高校科学研究计划资助项目(XJEDU2004E02;XJEDU2006I10)

摘　　要：混沌神经网络能有效地解决函数优化问题。通过把sigmoid函数转化为墨西哥帽小波函数,而单一化退火因子函数被分段指数模拟退火函数所取代,提出了一种新型的混沌神经网络。与传统的混沌神经网络相比,该网络具有更强的全局寻优能力。仿真结果表明,小波混沌神经网络在搜索全局最优解的速度和精确度上都明显优于传统的混沌神经网络。Chaotic neural network can solve function optimization problems effectively, A wavelet chaotic neural network was proposed by transferring sigmoid function to Mexican hat wavelet function and the single-parameter annealing function being replaced by subsection exponential simulated annealing functions. In contrast to the conventional chaotic neural network, it has a much higher ability to find the global optimal value. The simulation results show that on rapidity and accuracy of searching for the globally optimal solution, this wavelet chaotic neural network is obviously superior to conventional chaotic neural network models.

关 键 词：小波混沌神经网络 墨西哥帽小波函数 指数模拟退火函数 函数优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]