论文部分内容阅读
该文结合终点吸引子与罚函数方法,构造了求解满足约束条件的最小向量范数问题的终点吸引子神经网络模型,所提出的方法克服了惩罚因子须取充分大的要求,使得网络易于收敛于稳定状态.利用正则化方法解决了求解向量范数极小化中的病态问题,从而得到了一个正则化神经网络.所有的网络均给出了电路结构图,我们的方法有利于VLSI实现及其实时求解.用具体例子进行了模拟实验,实验结果说明了方法的正确性与有效性.“,”In this paper, terminal attractors neural networks for solving minimal vector norm problem subjected to equality constraints are constructed by means of terminal attractors and penalty function method. The proposed method need not take sufficiently large penalty factors. And network easity converge to stable state. Ⅲ-conditional problems for solving minimal vector norm are solved by use of regularization methods. Therefore, A regularization neural network is obtained. Circuit architecture figure for all networks are given Our methods are best suited for VLSI implementations and real-time soluution. computer simulationsare Underway by use of specific examples. Simulation results demonstrate correctress and effectiveness of our methods.