计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (23): 75-76.DOI: 10.3778/j.issn.1002-8331.2008.23.023

• 理论研究 • 上一篇    下一篇

一种求解线性约束的非线性规划神经网络方法

申 芸,吕咏梅,周永权   

  1. 广西民族大学 数学与计算机科学学院,南宁 530006
  • 收稿日期:2007-07-23 修回日期:2008-01-02 出版日期:2008-08-11 发布日期:2008-08-11
  • 通讯作者: 申 芸

Neural network for nonlinear programming with linear constraints

SHEN Yun,LV Yong-mei,ZHOU Yong-quan   

  1. College of Mathematics and Computer,Guangxi University for Nationalities,Nanning 530006,China
  • Received:2007-07-23 Revised:2008-01-02 Online:2008-08-11 Published:2008-08-11
  • Contact: SHEN Yun

摘要: 针对线性约束的非线性规划的求解问题,利用罚函数求解优化问题的思想将其转化为二次凸规划,基于神经网络的结构特性,定义所需的能量函数,从而使网络收敛于唯一稳定点最终实现线性约束的非线性规划的求解。实验仿真结果表明,该方法是有效和正确的,且能推广到含参的非线性规划和多目标规划中去。

关键词: 线性约束, 非线性规划, 神经网络, 稳定点

Abstract: In this paper,an algorithm for solving the optimal solution of nonlinear programming with linear constraints is proposed.The problem is translated into convex quadratic programming basis of penalty function for solving optimal problem.Then the energy function which is content with converging on the stabilization point by the properties of neural network is defined.The result of the computer experiment shows that the proposed method is effective and correct.It can also solve parametric nonlinear programming and multi-objective programming.

Key words: linear constraints, nonlinear programming, neural network, stabilization point