网络设计中集中器定位问题的神经网络解决方法

A Neural Network Approach to Concentrator Location in Network Design

给出了用于解决网络设计中集中器定位问题的神经网络方法。这类容量受限的集中器定位问题是求最小总花费的NP完全问题 ,利用拉格朗日乘数法和惩罚函数构造神经网络动态系统的微分方程。采用专门的神经网络可以有效解决。模拟结果表明 ,这种神经网络方法有效可行 ,并能求出最优解或近似最优解。

In this paper,a neural network approach is made to the concentrator location problem(CLP)in the design of local access network The goal of this NP-complete problem is to minimize the total cost associated with capacitated concentrators and to apply the lagrangian method and the differential equation of the neural network dynamic system made of penalty functions In this paper,the problem is efficiently solved by the special neural network Several simulation results show that the neural network approach is desirable in finding better solutions or optimal solutions