Determining of an Optimal Maintenance Policy for Three State Machine Replacement Problem Using Dynamic Programming

In this article, we present a sequential sampling plan for a three-state machine replacement problem using dynamic programming model. We consider an application of the Bayesian Inferences in a machine replacement problem .The machine was studied at different states of good, medium and bad. Discount dynamic programming (DDP) was applied to solve the three-state machine replacement problem, mainly to provide a policy for maintenance by considering item quality and to determine an optimal threshold policy for maintenance in the finite time horizon. A decision tree based on the sequential sampling which included the decisions of renew, repair and do-nothing was implemented in order to achieve a threshold for making an optimized decision minimizing expected final cost. According to condition-based maintenance, where the point of defective item is placed in continuing sampling area, we decided to repair the machine or to continue sampling. A sensitivity analysis technique shows that the optimal policy can be very sensitive.