改进遗传神经网络控制混沌运动的研究  

STUDY ON CHAOS CONTROL WITH ONE IMPROVED GENETIC NEURAL NETWORK

作　　者：陈玲莉 谭宁[1] 黎红岗[1] 梁欧[1] 

机构地区：[1]西安交通大学航天航空学院,西安710049

出　　处：《动力学与控制学报》2009年第1期24-28,共5页Journal of Dynamics and Control

基　　金：国家自然科学基金资助项目(10602003)~~

摘　　要：用最大Lyapunov指数构造遗传算法中的适应度函数,通过遗传算法优化神经网络的权系数.根据所得到的适应度函数和权系数来构造遗传神经网络控制器,从而提高神经网络控制效果.对离散系统Logistic映射和连续系统Rossler方程、AFM(原子力显微镜)悬臂梁振动系统的混沌运动分别进行了仿真控制.数值实验结果表明本文改进的遗传神经网络控制方法对离散或者连续的混沌系统都能控制到低周期轨道上去,证明了算法的有效性.The author constructed the fitness functions of genetic algorithms with the maximum Lyapunov exponent, and optimized the weights of a neural network with the genetic algorithm. According to the fitness functions and the weights, a genetic algorithm neural network controller was designed, which improved the efficiency of neural network control. Numerical simulations were performed for the discrete system as Logistic mapping, continuous system as Rossler equations, and the chaotic transition of the AFM microcantilever. The results of numerical experiments indicated that the improved genetic neural network control method can control discrete and continuous chaos systems to the expected period orbit, and prove fhe reliability of the algorithm.

关 键 词：混沌控制 改进遗传算法 神经网络控制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] O242.23[自动化与计算机技术—控制科学与工程]