Application of Markov programming in discrete manufacturing control

This paper focuses on selecting the minimum cost control for a discrete manufacturing system. Information is available to the controller concerning both the process and the status of the incoming part. A Markov programming approach is utilized to explicitly consider the stochastic elements of the process. That is, the part characteristic is assumed to be transformed by the process according to a probability distribution that is unique for each mode of control. Selection of the minimum cost control is performed based on these distributions and information concerning the status of the incoming part. Specifically, the three scenarios assume that the controller is provided with: 1) no information, 2) the distribution of the quality characteristic among the potential values, and 3) complete information about each part. A model for each scenario is developed and an example is presented to further illustrate the concepts.