带有高斯扰动的混沌神经网络及应用

PDF下载HTML

Title:

A novel chaotic neural network with Gaussian disturbance and its application

作者:

许楠¹, 刘桂阳¹, 徐耀群²

1. 黑龙江八一农垦大学 信息技术学院, 黑龙江 大庆 163319;
2. 哈尔滨商业大学 系统工程研究所, 黑龙江 哈尔滨 150028

Author(s):

XU Nan¹, LIU Guiyang¹, XU Yaoqun²

1. College of Information Technology, Heilongjiang Bayi Agricultural University, Daqing 163319, China;
2. Institute of System Engineering, Harbin University of Commerce, Harbin 150028, China

关键词:

高斯函数; 扰动; 混沌神经网络; 内部状态

Keywords:

Gaussian function; Gaussian disturbance; chaotic neural network; inside state

分类号:

TP18

DOI:

10.3969/j.issn.1673-4785.201308013

摘要:

为考查新型混沌神经网络的抗干扰能力, 在Chen’s混沌神经网络模型的内部状态中加入高斯函数扰动项。通过分析该网络神经元的混沌特性, 说明了模拟退火参数及高斯函数的宽度参数对混沌行为的影响;分析了不同宽度值时的高斯曲线以及内部状态随迭代次数的变化情况, 说明了扰动强弱敏感依赖于该参数的取值。将带有高斯扰动的混沌神经网络模型应用于解决旅行商最短路径问题(TSP), 通过仿真实验, 说明若合理调整网络参数以及径向基宽度参数, 则该新型网络可以具有较高的抗扰动能力, 从而使得网络能够避免陷入局部极小点, 并以较快速度求得全局最优解。

Abstract:

A Gaussian function disturbance item is added into the internal state of Chen’s chaotic neural network for examining the anti-disturbance ability of the new chaotic neural network. The chaotic dynamics behavior of the single chaotic neuron is analyzed. The chaotic behavior is affected by the parameter of the simulated annealing and the width of the Gaussian function. The Gaussian curve and the change of the inside state are analyzed in different widths. The amount of disturbance depends on the width. This chaotic neural network with Gaussian disturbance is used to solve traveling salesman problem. The simulation results indicate that this network can avoid the limits of being trapped into the local minima and its capability of resisting the disturbance is perfect.