Optimization Of Supply Chain Planning Under Decentralized And Uncertain Environment

Decentralization and uncertainty are two key characters of the supply chainplanning. So it is important to research the optimization of supply chain planningunder decentralized and uncertain environment. The main contributions of thedissertation consist of the following four parts:1. For the closely-cooperative supply chain with a coordinator, adecentralized coordination method based on Lagrangian relaxation algorithmis presented. To accelerate the convergence of this method, a fuzzysubgradient algorithm, whose convergence is proved, is also presented toupdate Lagrangian multiplier. To obtain the feasible solution, a distributedforward heuristics algorithm is proposed. The computational experimentsshow that the method can solve the supply chain planning problem efficiently.Then, a method based on augmented Lagrangian relaxation is also given,which improves the solution with more computing time;2. For the loosely-cooperative supply chain without a coordinator, aasynchronous coordination method based on Lagrangian relaxation ispresented. The surrogate subgradient algorithm is used to update theLagrangian multipliers. To obtain the feasible solution, a distributedinteractive heuristics algorithm is proposed. The tests show that this methodcan solve the supply chain planning problem efficiently.3. For the fuzzy inventory problem, four kinds of problems areresearched as follows: 1) EOQ problem with fuzzy demand and fuzzy orderingcost;2) EOQ problem with fuzzy demand, fuzzy ordering cost and fuzzyinventory cost;3) EOQ problem with fuzzy defective rate and fuzzy orderingcost;4) Inventory problem with fuzzy demand and fuzzy lead time. For eachproblem, the best inventory policy is proved with triangle fuzzy number. Andfor the fourth problem, a heuristics algorithm is presented to get the best orderquantity and order point with general fuzzy number;4. For the production-distribution planning with fuzzy demand, a fuzzytwo-stage model is set up after analyzing the decision process of the problem.When demand is fuzzy, the possibility distribution of the optimal solution ispresented and proved. Then an alternative method is presented, which canalter the fuzzy problem into a crisp problem by the cut sets of fuzzy number.Then the optimal solution of the two-stage fuzzy problem can be obtainedeasily by solving the crisp problem. The computational experiments show thatthe method can solve the supply chain planning problem with fuzzy demandefficiently.