摘要
针对具有边界约束的凸二次规划问题,利用离散神经网络模型的建模原理,构造了一个新的离散投影神经网络模型。为了优化求解,运用李亚普诺夫稳定性理论,通过构造一个新的广义李亚普诺夫函数,得到了神经网络模型全局指数稳定的线性矩阵不等式条件。同时将结果的保守性与传统方法所得结果进行了比较。数值仿真表明,改进方法采用新条件,有求解方便,保守性弱的特点,具有较强的实用性。
In views of convex quadratic optimization with bound constrains,a new discrete time projective neural network is established by using the principle of mathematical model of discrete time neural networks model. Based on Lyapunov stability theory,by constructing a new general Lyapunov function,a new global and exponential stability condition is obtained in views of linear matrix inequality technique. Numerical simulations show that compared with some latest results obtained in the literature cited therein,the new conditions obtained in this paper can be easily solved and have less conservatism.
出处
《计算机仿真》
CSCD
北大核心
2010年第9期161-164,共4页
Computer Simulation
基金
教育部新世纪优秀人才支持计划(NCET-06-0811)
贵州财经学院博士项目(200702)
关键词
离散投影神经网络模型
全局指数稳定
二次最优化
线性矩阵不等式
数值仿真
Discrete projective neural networks model
Global exponential stability
Quadratic optimization
Linear matrix inequality
Numerical simulation