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

Abstract: 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 simulated annealing method and a simulation-based Particle Swarm Optimization method are designed and compared. The sensitive 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.

Key words: inventory control, stochastic stock-dependent demand, simulation-based evolutionary algorithm, Particle Swarm Optimization（PSO）, simulated annealing