一类求解退化凸二次规划的投影神经网络  

A Projection Neural Network for Solving the Degenerate Quadratic Program

作　　者：单彩虹[1] 武怀勤[1] 

机构地区：[1]燕山大学理学院,河北秦皇岛066004

出　　处：《五邑大学学报（自然科学版）》2009年第1期57-62,共6页Journal of Wuyi University(Natural Science Edition)

摘　　要：借助变分不等式和Kuhn-Tucker条件,构造了一类投影神经网络求解线性约束的退化凸二次规划问题.与已有的求解退化凸规划问题的神经网络系统相比,系统的适用范围更广;在理论方面,系统是全局收敛的;数值实例显示了所得结论的有效性和正确性.By making use of variational inequality and Kuhn-Tucker conditions, we develop a projection neural network for solving degenerate quadratic programming problems with general linear constraints. Compared with the existing neural networks for solving degenerate convex quadratic program, the proposed neural network has a wider domain for implementation. In the theoretical aspects, the proposed neural network is shown to have global convergence. Illustrative examples show that the proposed neural network is effective and correct.

关 键 词：退化凸二次规划 投影神经网络 变分不等式 K-T条件 全局收敛 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]