二次规划问题的新型时滞投影神经网络的稳定性分析

本文针对求解有等式约束的二次规划问题的时滞投影神经网络，提出一种所有神经元皆存在时滞情况的新型网络模型。利用Gronwall不等式和Halanay不等式，严格证明了所提网络模型的全局指数稳定性。研究结果表明，所提网络模型的解能够渐进收敛于原问题的最优解，具有计算迅速和精度较高等特点。数值仿真结果验证了所提网络模型的可行性与有效性。

Stability Analysis of Novel Delayed Projection Neural Networks for Quadratic Programming Problems

in view of delayed projection neural network used for sotving quadratic programming problems with equality constraints, a novet delayed projection neural network with time delay in all the neurons is proposed in this paper. Based on the Gronwall inequality and the Halanay inecluality, it is strictly proved that the model is globally exponential convergent. Moreover, the .research. shows that it can asymptotically converge to an exact optimal solution, which is faster in computation and of high accuracy. Simulation results demonstrate the feasibility and effectiveness of the model.