摘要
针对生鲜农产品零售商库存成本控制问题,将该问题转换为马尔可夫决策过程,引入三参数Weibull函数,描述生鲜农产品的损腐特征,并考虑过期、损腐、缺货、订货和持有等成本,从供应链视角建立生鲜农产品库存成本控制模型,使用深度强化学习中深度双Q网络(Double Deep Q Network,DDQN)优化订货,以控制库存总成本。实验结果表明,相比单周期随机型库存成本控制模型和固定订货量库存成本控制模型,DDQN模型的总成本分别降低约6%和11%,具有实际应用价值。
In this paper,we solve the inventory cost control problem of fresh agricultural retail by transforming it into a Markovian decision process.A three-parameter Weibull function is introduced to describe the spoilage characteristics of fresh agricultural products,and the costs of expiry,rot,out-of-stock,ordering and holding are considered.We establish a fresh agricultural product inventory cost control model from the perspective of supply chain,and use the Double Deep Q Network(DDQN) in deep reinforcement learning to optimize ordering to control the total inventory cost.The experimental results show that the total cost when using DDQN inventory cost control model is reduced by about 6% and 11% respectively compared with that when using the single-cycle stochastic inventory cost control model and the fixed order quantity inventory cost control model.
作者
李姣姣
何利力
郑军红
LI Jiaojiao;HE Lili;ZHENG Junhong(College of Computer Science and Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China)
出处
《智能计算机与应用》
2023年第10期60-64,72,共6页
Intelligent Computer and Applications
基金
浙江省重点研发计划(2022C01238)。
