变几何桁架机构位置分析的神经网络超混沌法研究

The Research of Hyper-chaotic Neural Network Method to Displacement Analysis of Variable Geometry Truss Manipulator

自然科学与工程中的许多问题都可以转化为非线性方程组的求解问题,牛顿迭代法是重要的一维及多维的迭代技术,其迭代本身对初始点非常敏感.通过消除暂态混沌神经元的模拟退火策略,产生了一种可以永久保持混沌搜索的混沌神经元,研究了由4个该混沌神经元全连接的混沌神经网络的拓扑结构,混沌神经网络中存在超混沌现象(具有3个正的李氏指数).应用神经网络超混沌系统产生牛顿迭代法的初始点,提出了基于神经网络超混沌的牛顿迭代法求解非线性方程组的新方法.变几何桁架机构综合实例表明该方法的正确性与有效性.图3,表1,参14.

Many questions in natural science and engineering are transformed into nonlinear equations to be found, Newton iterative method is an important technique to one dimensional and multidimensional variables and it＇s iterative process exhibits sensitive dependence on initial guess point. By eliminating the simulated annealing mechanism of transiently chaotic neuron, a kind of chaotic neuron, which can permanently sustain chaotic search, was presented. Based on the complete connection topology of chaotic neural network composed of the four chaotic neurons,hyper-chaos with three positive Lyapunov exponents exist in the chaotic neural network system A novel method to find all solutions based on utilizing hyper-chaofic neural network to obtain locate initial points to find all real solutions of the nonlinear questions was proposed. The numerical examples show that the method is correct and effective.