Apply Markov Decision Process with Variable Discrete Random Demand Problem

摘要

We describe a formulation of the dynamic lot-sizing single product problem when demand is discrete random and time variant. Assuming that the probability of each demand alternative is known for each period. The problem is difficult to solve if there is size increasing because of the large amount of alternative demands. It is required to take longer computational time to obtain optimal solution. The solutions are usually available for problems with a variable discrete random demand in a short period. This study was investigated the solution by applying the Markov decision process (MDP) and develop a linear model for the MDP to solve the problem. A case study is used for uncertain demand to determine fixed price. According to testing, this solution was able to solve the problem consisting short period and used with the problem of a variable discrete random demand over a long period. The MDP is not an integer programming model with short computational time. The results have been the conducted to solve a large problem.