基于霍普费尔德神经网络法的ATM计算机通信网络中虚通路的拓扑优化

为了求解ATM计算机通信网络中虚通路的拓扑优化问题，提出了一种基于霍普费尔德神经网络的虚通路拓扑优化算法，给出了神经网络的能量函数及其神经元的动态方程，通过大量的计算机仿真实验，获得了选择系统参数的经验公式，确定出虚通路的拓扑优化问题的系统参数，所得结果能够以较快的速度收敛到具有很高稳定性和有效性的最优解和近优解，并且与传统的最短路径路由法和最小负载路由法比较，本算法所得解的质量更好．这些研究结果对于解决宽带综合业务数字网及其他工程系统中大量出现的复杂优化问题具有重要的理论价值和广阔的应用前景．

Topological optimization of the virtual path in ATM computer communication networks based on the Hopfield neural network method

In order to solve the topological optimization problem of the virtual paths in the ATM com- puter communication networks,a topological optimization algorithm for the virtual paths based on Hop- field neural networks is presented,and the expression for the energy function and neuron dynamic equa- tion in the neural networks are given.With a good number of the computer simulation experiments,the experience formulas of selecting the system parameters are obtained,and the system parameters of the topology optimization problem for the virtual paths decided.They demonstrate thatthe outcomes of this algorithm can be more rapidly converged to the mostoptimal solutions or the near optimal solutions with very high stability and validity,and the quality of solutions obtained by the algorithm is better than that of the traditional shortest path routing method and the least loaded routing method.These results are of important theoretical value and broad application to the solution of complex optimization problems emerging in the broadband integrated services digital networks and other engineering systems.