节点文献
求解二次规划的一个递归新神经网络
A new recursive neural network for solving quadratic programs
【摘要】 针对一类不等式约束的二次规划问题,分别采用Lagrangian系数获得新模型和构造合适的Lyapunov函数的方法,求得对偶问题,通过代换得到初始问题的最优解,并证明新模型的全局稳定性.结果表明:新模型公式简单直观,计算量少,收敛速度快,且对初始问题有更精确的解,通过计算机仿真实验验证了算法的有效性.该成果对二次规划稳定性的研究和应用提供了理论依据.
【Abstract】 For a class of quadratic programming problems, inequality constraints Lagrangian coefficients were used respectively to get the new model, and constructed the suitable Lyapunov function method for the dual problem. The optimal solution of initial problem was obtained by substitution, and the global stability of the new model was proved. The results show that the new model formula is simple and intuitive with less amount of calculation, and fast convergence rate, and it has a more accurate solution to the initial problem. Through the computer simulation results, the effectiveness of the algorithm was verified. The study of the stability of the quadratic programming has certain reference and application value.
【Key words】 neural network; quadratic programming; differential equations; global optimization; dual problem; Lyapunov function; global asymptotic stability; Lipschitz condition;
- 【文献出处】 辽宁工程技术大学学报(自然科学版) ,Journal of Liaoning Technical University(Natural Science) , 编辑部邮箱 ,2014年01期
- 【分类号】O221.2
- 【被引频次】1
- 【下载频次】135