求解二次规划问题的离散时间神经网络的收敛性分析  

CONVERGENCE ANALYSIS OF DISCRETE-TIME NEURAL NETWORK FOR SOLVING QUADRATIC PROGRAMMING PROBLEMS

作　　者：路洋[1] 李德伟[1] 席裕庚[1] 卢建波[1] 

机构地区：[1]上海交通大学自动化系系统控制与信息处理教育部重点实验室,上海200240

出　　处：《系统科学与数学》2012年第11期1343-1353,共11页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(60934007;61074060;61104078);中国博士后科学基金(201003272);上海市教委科研创新项目(11CXY08)资助课题

摘　　要：对求解二次规划问题的离散时间神经网络的收敛性进行了分析,通过选取适当的李雅普诺夫函数给出了网络全局收敛的充分条件,并在该条件下研究了网络的收敛速度,分别对问题的不等式约束左矩阵行满秩和非行满秩的情况进行了讨论,得到了在上述充分条件下对于不等式约束左矩阵行满秩和非行满秩的问题均有网络指数收敛的结论,通过仿真验证了结论的正确性.The convergence property of discrete-time neurM network for quadratic pro- gramming is analyzed. By choosing a proper Lyapunov function, a sufficient condition for global convergence is obtained. The convergence rate under the condition is also investigated through a in-depth discussion about full-row-rank inequality constraint left matrix condition and non-full-row-rank inequality constraint left matrix condition, respectively, and the expo- nential convergence property for both full-row rank and non-full-row rank inequality constraint left matrix conditions under the mentioned sufficient condition is proved. Simulation result verifies the validity of the theoretical results obtained in this paper.

关 键 词：二次规划 神经网络 离散时间 收敛性 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] O221[自动化与计算机技术—控制科学与工程]