Closed-loop Scheduling
Chemical production scheduling, and scheduling in general, is an essential task which occurs in a variety of industrial applications. In the last few decades, research within the scheduling community has led to significantly improved optimization algorithms and problem modeling capabilities. However, the development of algorithms and theory for reactive, or closed-loop, scheduling has received considerably less attention.
Model predictive control (MPC) provides a natural framework and set of theoretical properties to understand the behavior of closed-loop systems. Recently, standard scheduling problem formulations have been represented in a state-space framework (Subramanian, Maravelias, and Rawlings, 2012), properties concerning economic MPC have been analyzed for periodic systems (Angeli, Amrit, and Rawlings, 2012), and MPC has been extended to include discrete actuators (Rawlings and Risbeck, 2017). These are all essential steps towards applying the results of MPC towards scheduling problems. To further this goal, I am investigating algorithmic and theoretical results from control theory which can be used to improve and analyze the behavior of closed-loop scheduling.
This project is in collaboration with Johnson Controls, Inc.