Research On Model And Algorithm Of Stochastic Joint Replenishment Problem

Joint replenishment problem is a very significant problem in the real production/inventory control theory and practice. When a group of goods are supplied by the same supplier (or from the same supply place), or when a group of goods are transported by a transport vehicle (such as automobile, steamship, airplane), we have to confront the problem how to coordinate and arrange reasonable order to minimize the cost. Most of previous literatures on the replenishment problem were confined to deterministic demands of deterministic inventory system. In view of some practical situations of enterprises in our country, for example, goods order may be delayed at any time. When demand and order lead time is stochastic, we have to analyze and model according to stochastic inventory system.Firstly, stochastic replenishment inventory problem with capital and storage capacity constraints in the environment of a single location-multiple product is considered. Based on the periodic replenishment policy, the method of making stochastic problem confirmed is used; stochastic demand is transformed to deterministic demand and this error is corrected by deviation factor. Then, stochastic optimal model is constructed to minimize the sum of ordering cost and inventory (including safety inventory) holding cost and a genetic algorithm (GA) is proposed to solve the model. And then this algorithm is verified by simulation results.Secondly, stochastic joint replenishment problem that is discussed in the dissertation not only can be applied to a single location-multiple product environment, but also to one product-multiple locations environment. On the assumption that demand and order lead time of every retailer are stochastic, we research stochastic joint replenishment problem under the one product-multiple locations environment. In order to reach minimal inventory cost per unit time, a (s, S) model based on stochastic demands and lead time is advanced, a modified genetic algorithm based on minimal gene segment coding, two generations competition and adaptive selection are developed to solve the problem, a Monte Carlo method is presented to compute adaptive values, which significantly, improve capital management and control of every retailer. And an example data is introduced to verify the result.Finally, an inventory system is developed and the inventory decision of stochastic joint replenishment problem under a single location-multiple product environment is applied to verify theoretical results. The feasibility and application value of joint replenishment problem model is displayed fully.