基于蚁群算法的供应链滚动优化决策方法  

Receding-horizon decision-making of supply chain based on ant colony optimization algorithm

作　　者：杨春节[1] 何川[1] 

机构地区：[1]浙江大学控制科学与工程学系,浙江杭州310027

出　　处：《计算机集成制造系统》2010年第1期133-139,173,共8页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(60674086)~~

摘　　要：针对钢铁供应链最终客户的需求不确定性,提出了一种将蚁群算法与滚动优化算法相结合的供应链优化决策方法,旨在满足最终客户不确定性需求的同时降低成本。通过实施滚动优化策略,来减少需求不确定性导致的决策失误。在优化模型中考虑了采购规模和生产规模对单位成本的非线性影响。为求解滚动优化中的非线性优化问题,通过将成本等效为路径的长度,将决策变量的候选解等效为城市,从而把决策优化问题转化为蚁群路径寻优问题。在每次静态优化中,优化算法根据历史数据和反馈信息来确定优化决策变量。针对一个包括供应商、生产商、零售商和最终客户的供应链对象进行了仿真研究,结果表明了所提方法在克服需求不确定和模型非线性方面的有效性。Aiming at uncertain requirements of the end customer in iron steel supply chain,a supply chain optimization decision method by integrating ant colony algorithm and receding-horizon algorithm was presented to satisfy uncertain requirements and reduce the cost at the same time.Implementation of receding-horizon algorithm reduced wrong decisions caused by uncertain requirements.The nonlinear influence of purchasing scale and production scale were all considered in this optimization model.To solve the nonlinear optimization problem in receding-horizon optimization,the decision optimization problem was converted to ant colony route optimization problem in which the costs were equivalent to the length of routes and the candidates of decision variables were equivalent to the cities.In each static optimization process,the optimization decision variables were determined by historical data and feedback information.Simulation cases were conducted in a supply chain including the supplier,the manufacturer,the retailer and the end customer.Results demonstrated the proposed method＇s good performance in satisfying uncertain requirement and the model nonlinearity.

关 键 词：供应链管理 蚁群算法 滚动优化 决策 钢铁企业 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]