八面体变几何桁架机构综合的神经网络超混沌牛顿迭代法研究

Research on the hyper-chaotic Newton iteration method of neural network for the synthesis of octahedron variable geometry truss mechanism

神经网络是高度复杂的非线性动力系统,存在着混沌现象。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个该混沌神经元连接的单向循环混沌神经网络拓扑结构和混沌神经网络中存在超混沌现象。应用神经网络超混沌系统产生牛顿迭代法的初始点,首次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。八面体变几何桁架机构综合实例表明了该方法的正确性与有效性。

Neural network is a highly complicated nonlinear dynamic system, and there exists the chaos phenomena. By eliminating the simulated annealing strategy of the transient chaos nerve cell, a kind of chaotic nerve cell that could permanently maintain the chaotic search was engendered. The topological structure of unidirectional circulating chaos neural network connected by those 4 chaos nerve cells and the hyper-chaotic phenomena existed in the chaos neural network were studied. The initial point of Newton iteration method was engendered by the use of neural network hyperchaotic system. This new method for solving the nonlinear equation set based on the Newton iteration method of neural network hyperchaos was put forward for the first time. The correctness and validity of this method were made clear by the mechanism synthesis of the octahedron variable geometry truss.