摘要
Hopfield 网络(HNN)是一种有效的优化模型,但存在易收敛到非法解或局部极小以及对模型参数与初值依赖性强的缺点。旅行商问题(TSP)是研究算法性能的典型算例,通过对其进行计算机仿真优化,分析归纳了HNN 模型存在缺点的原因,总结并提出若干改进方法与思想。同时,针对TSP问题的工程背景提出了若干发展性研究内容与方法。
Hopfield neural network (HNN) is an efficient optimization model, but it is easy to be trapped in local minima and illegal solutions and very susceptible to initial conditions. Through simulations with typical traveling salesman problem (TSP), some drawbacks of HNN are analyzed, and several improvements are summed up and proposed. Moreover, with respect to the powerful engineering background of TSP, some improving research and approaches are presented.
出处
《控制与决策》
EI
CSCD
北大核心
1999年第6期669-674,共6页
Control and Decision
基金
国家自然科学基金
国家攀登计划基金