随机需求条件下库存管理模型及灵敏度分析  

Inventory management model and sensitivity analysis for stochastic demand

作　　者：曹黄金[1] 柴跃廷[1] 刘义[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《清华大学学报（自然科学版）》2012年第11期1513-1517,共5页Journal of Tsinghua University(Science and Technology)

基　　金：国家"十一五"科技支撑计划项目(2011BAH18B02)

摘　　要：该文研究随机需求条件下参数实际值偏离预期值对库存系统造成的影响。在需求为Ornstein-Uhlenbeck过程的假设下,采用系统动力学方法将库存系统表示成具有常参数的随机微分模型。定义了状态对参数扰动的随机灵敏度,并采用直接微分法和伴随变量法得到了其评估算法。通过某打印机销售商某型号产品的库存管理仿真,得到了库存状态对订单处理时间、运输时间及调整系数扰动的灵敏度。这些结果有助于深刻理解库存系统行为,并可依此为管理者提供运作建议。This study analyzes the effect of differences between estimated and actual values on inventory systems with stochastic demand. Assuming that the demand follows an Ornstein-Uhlenbeck process, the inventory system is represented as a stochastic differential equation model with constant parameters using system dynamics. Then, the stochastic sensitivity of the states to small variations in individual parameters is evaluated by direct differentiation and adjoint variable methods. Numerical simulations for product inventory management of a printer vendor give the sensitivities of the inventory states to perturbations of the order processing time, transit time and adjustment factor. These results provide further understanding of the system behavior and provide operational recommendations for managers.

关 键 词：库存管理 随机需求 灵敏度分析 直接微分法 伴随变量法 

分 类 号：F253.4[经济管理—国民经济]