Two Supply Chain Optimal Models And Optimization Algorithms

A multi-echelon inventory stochastic model of supply chain was proposed.Maximizing the gross profit of supply chain is regarded as an objective of the model. Customer's demand and short goods that are supplied by lower nodes are stochastic. To reflect the influence of stochastic, quantity of short goods that are supplied by lower nodes on higher nodes,the concept of "effective supply" was introduced. Optimization method and analysis of model in a kind of special situation were provided. The result of analysis shows that model's optimal solve is affected by the uncertainty of short goods that are supplied by lower nodes.A 0-1 mix integer programming model of echelon supply chain was proposed. The model is based on the contradiction between customer satisfaction and enterprise profit. In the model supply and demand amount between every adjacent echelon nodes and customer's expectation satisfaction to various kinds of products are regarded as decision variables, the gross profit of supply chain is regarded as the objective of function. The model is a two layers optimal question. Firstly some nodal enterprises are selected from candidate nodes to join supply chain, then supply and demand amount between nodal enterprises of adjacent layer and customer's expectation satisfaction to the products are ascertained.Hybrid genetic algorithm for the model was provided ,and the optimal model's correctness and algorithm's effectiveness were confirmed by numerical experiments.A kind of hybrid genetic algorithm for integer programming was proposed.To keep the efficiency of the algorithm from dropping largely,a kind of new thought was used to combine genetic algorithm and simulated annealing algorithm in the algorithm. To guarantee algorithm converge to optimum solve and keep the efficiency of algorithm from dropping, a kind of new "nature select" was joined algorithm.In the algorithm several individuals were generated by a pair of parents.to increase population individuals variety and algorithm local search ability. The algorithm's convergence was proved by theory of Markov chain, and the effectiveness of the algorithm was confirmed by numerical experiments.