动态系统影响优化分析

Influence-optimization analysis for dynamic systems

管理者通常需要确定系统参数及决策变量,以达成预期目标。但优化参数较多时,往往导致求解精度较低,甚至无法求解。为提高优化效果,在定义影响度的基础上提出了影响优化分析方法。依据影响度结果,从众多系统参数中选取那些对系统目标影响较大的参数。将原优化问题转化为非线性规划问题,引入遗传算法优化控制序列和所选参数。以库存系统为例进行数值仿真计算,得到了不同预期、需求条件下的订单处理时间和订货规律。该方法有效减少了优化变量个数,可更准确地为管理者制定运作计划提供依据。

Managers often need to determine the system parameters and decision variables to achieve their desired goals.However,excessive parameters may reduce the accuracy and lead to no solutions.An influence-optimization analysis was developed based on the influence degree to identify parameters that have significant impact on the system goals.The original optimization problem is then converted into a nonlinear programming problem,with the control sequences and selected parameters obtained using a genetic algorithm.Simulations of an inventory system accurately predict order-processing time and ordering rules for different expectations and demand conditions.The number of optimization variables can be reduced using this method to improve the accuracy of operational plans.