Application of model predictive control to batch processes

Abstract

Model predictive control is applied to batch recipe synthesis scheduling problems. Due to the complexity of this class of problems, a state explosion situation exists where an exponential number of possible combinations of tasks exist in the solution set. In order to overcome this situation the batch processes are modeled discretely with Petri nets. This discrete model can be used to create a coverability tree. The coverability tree can be used to eliminate combinations of tasks, which are not safe or not useful to the process. As a result, the number of possible combinations of batch tasks is reduced to the point that the model predictive control algorithm can be used as a batch recipe synthesis package. This technique is applied to a batch process and compared with another technique. These two techniques yield similar results.