数据驱动下考虑风险规避的闭环供应链分布式鲁棒优化设计  

Data-driven Distributionally Robust Optimization Design for Closed-loop Supply Chain Considering Risk Aversion

作　　者：胡耀 李晓萍[1] HU Yao;LI Xiao-ping(School of Economics and Management,Jiangsu University of Science and Technolohy,Zhenjiang 212100,China)

机构地区：[1]江苏科技大学经济管理学院,江苏镇江212100

出　　处：《物流工程与管理》2024年第5期36-42,70,共8页Logistics Engineering and Management

基　　金：国家自然科学基金青年项目(71902074)。

摘　　要：随着可持续发展理念得到社会广泛认同,企业开始逐步关注供应链运作中资源的回收利用。为了合理规划资源,构建了以供应链总成本为目标,考虑多主体参与的闭环供应链混合整数规划模型。进一步考虑了采购价格不确定,结合分布式鲁棒优化方法,使用数据驱动模糊集模拟不确定参数的真实分布,并融入均值-CVaR方法衡量决策者的风险规避特征,通过对偶理论将模型转化为可处理的线性问题。随后在Python中调用Gurobi求解器对模型求解,结果显示分布式鲁棒优化可以较好地处理决策过程中不确定参数的影响,并且模型可以为不同决策偏好以及不同目标预算下的决策提供支持,最后通过灵敏度分析再次验证了模型的实用性。With the widespread recognition of the concept of sustainable development in society,enterprises are gradually paying attention to the recycling and utilization of resources in supply chain operations.To plan resources reasonably,a mixed integer programming model for closed-loop supply chain considering the participation of multiple entities was constructed with the total cost of the supply chain as the goal.Further considering the purchase price uncertainty,combined with distributionally robust optimization methods,data-driven fuzzy sets were used to simulate the true distribution of uncertain parameters,the mean-CVaR method was incorporated to measure the risk aversion characteristics of decision-makers,and the model was transformed into a manageable linear problem through duality theory.Then,the Gurobi solver was called in Python to solve the model,and the results show that distributionally robust optimization can effectively handle the impact of uncertain parameters in the decision-making process,and the model can provide support for decisions under different decision preferences and target budgets.Finally,the practicality of the model was verified again through sensitivity analysis.

关 键 词：采购价格不确定 分布式鲁棒优化 均值-CVAR 

分 类 号：F274[经济管理—企业管理]