需求随机依赖库存环境下的订货仿真优化模型  

Simulation and optimization of lot-sizing model with stochastic stock-dependent demand

作　　者：欧剑[1] 闵杰[1] 

机构地区：[1]安徽建筑大学数理系,合肥230601

出　　处：《计算机工程与应用》2014年第9期231-236,共6页Computer Engineering and Applications

基　　金：国家自然科学基金(No.71101002);安徽省高校自然科学研究项目(No.KJ2013B060);安徽建筑大学青年科研专项

摘　　要：需求的随机性和依赖库存性是库存问题的特点之一,在需求以泊松分布的形式随机依赖库存的条件下讨论了(Q,T)型库存控制问题。为了评估库存控制策略的平均盈利水平,建立了该库存问题的离散事件系统仿真模型,设计了一种基于仿真的种群重叠、遗传操作非重叠的进化算法,用以优化库存控制策略,类似设计了基于仿真的模拟退火和粒子群优化算法进行比较。通过实例分析了不同参数的变化对模型最优解的影响,灵敏度分析表明需求依赖库存效应越明显时,利润水平越高,最优订货策略越倾向于高库存、短周期和现货销售。仿真实例说明了基于仿真的优化算法的可行性、有效性。Stochastic demand and inventory-dependent demand are the characters of inventory control problems. By assuming that the demand of items obeys Poisson distribution, the（Q,T）inventory control model is discussed on condition of stock-dependent demand. A discrete event system simulation model of this inventory system is built to evaluate the average profit of this inventory control policy, and an improved simulation-based evolutionary algorithm with overlapping population and non-overlap genetic operators is designed to optimize the inventory control policy. Similarly, a simulation-based simu-lated annealing method and a simulation-based Particle Swarm Optimization method are designed and compared. The sen-sitive analyses of parameters show that the larger this dependency of demand on stocks the higher the profits level, and the optimal ordering policy should be higher-stock, shorter-cycle and stock sales. The simulation example shows that the simulation-based optimization algorithm is feasible and effective.

关 键 词：库存控制 需求随机依赖库存 基于仿真的进化计算 粒子群优化 模拟退火 

分 类 号：O227[理学—运筹学与控制论]