Research On Agricultural Supply Chain Network Optimization Under Demand Uncertainty

Our country has been vigorously gave support to the development of agriculture in strategy.High standards and strict requirements for agri-products from consumers makes the planning ofthe supply chain even more important, especially for the fresh products. Rapid development ofscience, technology, economic and so many networks of information have been changingcompetitive environment that agricultural operators always face. Especially, great variation ofconsumers’ demands for agri-products come out that the momentum of agricultural productsmarket competition is increasingly fierce. What agricultural enterprises need to think is how canthey stand out and make profits. As a result, reasonably optimizing the structure of supply chainnetwork has been pushed on agenda.The text analyzes reflections, sources and impacts of supply chain network, expoundsprinciples and steps during supply chain network optimization. Meantime, totally consider thetotal cost of suppliers, operation, transportation and wastage based on time-price parameter offresh agri-products under stochastic demands. And build up a single target chance-constrainedprogramming model with fuzzy parameters for a supply chain network which is simple source,single-product, multi-vendor, multi-stage and multi-user. According to fuzzy probability theoryin the field of mathematics, in the model, objective function and constraint conditions whichcontain the fuzzy parameters are respectively transformed into clear equivalence class, and thenthe uncertainty model becomes deterministic model.Considering about the complexity of the model, Lingo and Hybrid particle swarmoptimization algorithm are adopted to solve it. First of all, using Lingo to solve the MINLPproblem directly, and then design an algorithm based on PSO and DE, which is encoded withreal number to figure out the model. State the operations made up with encoding for thealgorithm, the initialization of population, design of fitness function, the crossover, mutation andselection. To get a solution for the model, the paper chooses Matlab software which isprogramming with PSO-DE algorithm. At last, it verifies the feasibility of the model by usingthis two methods through an example of fresh agricultural products supply chain network.Compared with the results，it turns out that Lingo is more convenient and correct than HPSO.Stochastic demands with the optimization of supply chain network for agri-products canbetter reflect when the real demands is uncertain, how to deal with facility location and reasonable distribution resources. Therefore, reasonable optimization of the agricultural productssupply chain network can effectively reduce the waste on resources and repeated operations,improve customer service levels, lower total cost of supply chain, provide some help and supportfor agricultural enterprises on decision-making.