一类连续随机型流量网络中流量分配的多目标优化问题

A class of multi-objective optimization problem on flow allocation in continuous stochastic-flow network

假设随机型流量网络的容量为连续型随机变量,为简化随机多目标规划期望值模型,采用最小路集的概念进行流量分配,消去了flow-conservation约束,并将原始问题分解为两步进行解决。利用多目标遗传算法对简化后的模型进行求解。实例验证结果表明,该算法有很好的通用性,能够很好地解决随机型流量网络上的流量控制分配问题。

The capacity of stochastic-flow network was supposed to be a continuous random variable. In order to predigest the expectation model of stochastic multi-objective planning, flow was allocated according to the conception of minimal path. So the flow-conservation constraints were eliminated and the original problem was decomposed to two tasks. The predigest model was solved by a multi-objective genetic algorithm. The tested results show that this algorithm is compatible and can commendably solve the problem of flow allocation in a stochastic-flow network.