Reinforcement Learning Algorithm Of Fuzzy Joint Replenishment Problem In Supply Chain

Now researchers are quite concerned about inventory issues, which inventorymanagement is an important part of supply chain management. Joint replenishmentproblem is a research problem under the environment in a wide range of items forinventory controlled. It refers to the order under joint decision-making from thesupply point ordered a variety of items, replenish inventory, using the appropriatedecision makers to judge, and thus meet the needs of the premise to pay theminimum total cost per unit time. we use simulation ways to address the jointsupplementary model which’s needs are fuzzy variables. We use reinforcementlearning strategy to adjusted varieties of goods’ period to sync, to reduce cycle timeof check inventory levels and bring the considerable cost savings.First of all, the study on the joint replenishment model of stochastic demandwith the shortage loss has been based on periodic replenishment strategy analyzingthe loss model caused by out of stock with different rate of deterioration and takingminimum of ordering costs, inventory holding costs and the loss of out of stock asthe objective function to establish the mathematical model. By studying the problemof fuzzy model simplification, we can get a fuzzy model as the objective functionand the mathematical model by learning algorithm for processing, the objectivefunction is minimize ordering costs.Secondly, article gives initial space, state space, transfer function setting, etc.The Q was calculated with the system behavior and transition probabilities throughsemi-Markov decision chain. Then, article shows a program to solve JRP in detail,that using reinforcement learning algorithm to solve the constrained jointsupplementary question so specific issues. And given numerical simulation examplesillustrate the effectiveness.Finally, the algorithm and model have been applied in the distribution storagesystem to verify the feasibility of the algorithm. This system consists of five parts: commodity management system, purchasing management system, transportationsystem, store management system and basic support system. From that, the authorhas obtained the best ordering cycle and the best ordering costs, which clearlydemonstrates the excellent application of algorithm in this system.