Abstract: In order to resolve the optimal inventory control problem with random demands and lead time, a heuristic Dynamic Response Particle Swarm Optimization（DRPSO） is presented. The dynamic nonlinear optimal model of Optimal Inventory Control Policy（OICP） is established in supply chain system, and assuming that customer demand and lead time are linear and Gaussian random variable respectively, based on which, the actual process of economic operators is simulated by setting different variations, increments and frequency. On the basis of standard PSO algorithm, the adaptive mutation probability and the response mode of dynamic updating are introduced to improve the adaptability of particles for the dynamic environment. To simulate the uncertain supply chain environment, two types of goal movement with various uncertain demands and lead-time are investigated, which shows to be effective in locating the changing best order quantity and recorder point.

Key words: random demands, random lead time, optimal inventory control policy, real-time track, dynamic response particle swarm optimization

摘要： 针对随机需求提前期环境下的库存管理问题，提出一种启发式动态响应算法求取最优订货策略。建立最优订货策略的动态非线性优化模型，并设定客户需求和订货提前期分别为线性和高斯随机变量，通过变化形式、步长和变化频率的不同模拟实际经济运营过程；在微粒群寻优过程中引入柔性变异概率及动态更新响应方式，使微粒具有感知外界环境变化及对变化的响应能力，提高算法对复杂动态系统环境变化的适应性。实证分析结果证明了所提方法对最优订货量实时变化的动态响应能力。

关键词: 随机需求, 随机提前期, 最优库存控制, 实时追踪, 动态响应微粒群算法