Study On Multi-commodity Flow Supply Chain Network Model For Multi-criteria With Random Demand

With the development of information technology and the emergence of global economic integration, the competition among enterprises is intensified increasingly, the customers'level of consumption is increasing, and the uncertainty of market demand increases greatly. The decision-makers located in different levels pursue not only the maximum of their own interests, but also the shortest transportation time and the maximum of service level. Therefore, the study on multi-commodity flow supply chain network model for multi-criteria with random demand has certain theoretical and practical significance.Firstly, this dissertation systematically and comprehensively makes literature review of supply chain network equilibrium model, throughout the world. Then it analyzes the main ideas and methodology of existing supply chain network equilibrium models. Further, the game theory, as well as variational inequalities theory and spatial price equilibrium theory are introduced, so as to lay the theoretical foundation for follow-up study.Secondly, it presents a multi-commodity flow supply chain network model with multiple tiers of decision-makers, consisting, respectively, of manufacturers, distributors and retailers, who can compete within a tier but may cooperate between tiers. It considers multi-criteria decision-making for both the manufacturers and the distributors whereas the retailers are subject to decision-making with random demand. At the same time, it also considers the retailers can buy products from the manufacturers directly. The equilibrium models of each level and the whole network are developed by variational inequality method, along with their equilibrium conditions and economic interpretations.At last, the algorithm of variational inequality is studied. It comparatively analyzes the modified projection algorithm, along with discrete time algorithm and simulated annealing algorithm chiefly, and gives the basic idea of the algorithms and calculation steps, along with convergence analysis. The same numerical example is calculated by the modified projection algorithm and discrete-time algorithm combined with simulated annealing algorithm, respectively. The results of the numerical example prove the rationality of the model.