论文部分内容阅读
本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, the terminal attractor neural network model for solving the minimal vector norm problem that satisfies the constraints is constructed by combining the end point attractor and the penalty function. The proposed method overcomes the requirement that the penalty factor needs to be sufficiently large to make the network easy to converge to stable state. The regularization method is used to solve the ill-posed problems in the minimization of the norm of the solution vector, and a regularized neural network is obtained. All networks are given a circuit diagram, our method is conducive to VLSI implementation and its real-time solution. The simulation experiments are carried out with concrete examples. The experimental results show the correctness and validity of the method.