价格敏感型供应链网络规划模型及其混合遗传算法

Network Planning Model and its Hybrid Genetic Algorithm for the Price-Sensitive Supply Chain

针对价格敏感型供应链中,供应价格与批量之间通常具有非线性的函数关系,建立了采购、需求和物流服务环节均具有价格敏感性的多供应商、多配送中心和多零售商的供应链网络规划整数非线性规划INLP模型,设计解决此NP-hard问题的混合遗传算法,以启发式算法修复进化过程中出现的非法染色体,提高算法寻优速度,通过算例及与SAS/OR模块运行结果的比较,验证该算法具有很强的稳定性和高效性。

In a price-sensitive supply chain,each business partner as independent profit unit defines the relationship between batch order and ordering price based on members＇ profit preferences and resource conditions.The ordering quality of products is strongly related to a member＇s price decision-making processes for purchasing,supplying,production,logistics and selling activities.The operating cost and profit of a supply chain is different from that of total product quantity and distribution.Planning can reduce operating cost and improve profitability and market competitiveness for a price-sensitive supply chain.An effective planning can further lower resource consumption and make contributions to sustainable development of society and economy. This research studies a four-stage price-sensitive supply chain consisting of multiple suppliers,one manufacturer,multiple distribution centers and multiple retailers.A manufacturer needs to make decisions on supply chain network planning activities,such as supplier selection for parts and logistics support,sourcing from different suppliers,and distribution and selling of final products via different retailers.The network planning problem is complicated and requires supply chain managers apply a good decision-making algorithm to obtain a satisfactory solution on a timely basis. We first build an integer non-linear planning（INLP） model to represent network planning problems.We then use the heuristic repair method embedded with the hybrid genetic algorithm（HGA） to resolve the model.Finally,we use simulations to verify the model validity,as well as the stability and efficiency of the algorithm.It is essential to build a general structure model and mathematical model for optimizing supply chain planning due to the multi-forms of actual supply chains. In the first part,we describe the planning problem in detail,and design a general structure model for supply chains to represent the relationships among the members.We also build an INLP model to calculate price sensitivity of supply chain networking planning by including all quantity constraints.The values of decision variables increase exponentially as the number of nodes and node capacity increase.In the second part,a hybrid genetic algorithm is proposed to solve the INLP model.HGA techniques used in the study include integer encoding,heuristic repair strategy,representation of fitness function,and the selection of genetic operators.The primary focus of the algorithm is the heuristic repair strategy with price sorting rules.The strategy can not only ensure the validity of chromosomes,but also guarantee the lower cost of the chromosomes.In the third part,an example is simulated to check the validity and practicability of INLP model and HGA.Simulation results are compared to SAS/OR testing and show that our proposed algorithm can effectively resolve complex supply chain planning problems. Price-sensitive supply chain structures and pricing mechanisms are diverse.The network planning INLP model with the hybrid genetic algorithm can obtain a satisfactory planning solution for a price-sensitive supply chain.