基于映射矩阵的多目标交换式以太网拓扑优化  

Multi-objective Optimization of Switched Ethernet Topology Based on Mapping Matrix

作　　者：李杰林[1] 陈明[2] 陈德基[3] 

机构地区：[1]同济大学机械与能源工程学院,上海201804 [2]同济大学中德工程学院,上海201804 [3]同济大学电子与信息工程学院,上海201804

出　　处：《四川大学学报（工程科学版）》2016年第4期167-174,共8页Journal of Sichuan University (Engineering Science Edition)

基　　金：国家科技部重大专项子项资助项目(2009ZX04014-103);上海市经济与信息化委员会资助项目(11XI-07);上海市科学技术委员会专项基金资助项目(11DZ1121000)

摘　　要：针对工业现场交换式以太网网络传输时延过长、子网间负载不平衡的问题,在第二代非支配排序遗传算法的基础上,提出基于0-1映射矩阵的多目标交换式以太网拓扑优化方法。该方法利用网络拓扑0-1映射矩阵所具有的特殊性质构建多目标优化数学模型,采用类似无性生殖的单亲多点交叉杂交与双变异策略实现种群进化,进而得到交换式以太网网络拓扑映射矩阵的离散Pareto前沿。基准测试与仿真实例结果表明,网络拓扑结构优化后交换机子网间通信负载降低,局部网段内出现信息拥堵或功率闲置的几率减小,以太网的网络传输性能得到提高。Considering the low communication performance such as long time delay and imbalance problem of industrial switched Ethernet network, a multi-objective optimization method based on 0 - 1 mapping matrix and Non-dominated sorting genetic algorithm-Ⅱ was proposed. In this method, a muhi-objective optimization model was constructed by means of the special characteristics of 0 - 1 mapping matrix, multi-point asexual crossover and double mutation strategies were utilized to achieve population evolution, and then the corresponding discrete Pareto front of the switched Ethernet topology structure mapping matrix was obtained. Benchmark test and simulation results showed that the communication load between sub networks was dramatically reduced, and the possibility of information congestion or idle power in local networks was decreased after the network structure optimization.

关 键 词：以太网 网络拓扑 多目标优化 映射矩阵 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP393[自动化与计算机技术—控制科学与工程]