Mathematical optimization methods have been developed to a vast variety of complex problems in the field of process systems engineering (e.g., the scheduling of chemical batch processes). However, using these methods in online scheduling is hindered by the stochastic nature of the processes and prohibitively long solution times when optimized over long time horizons. The following questions are raised in online scheduling: when to trigger a new rescheduling procedure, how far ahead to schedule, what optimization strategy to use, and how much computation resources to allocate. The RELOOP project investigates an approach where reinforcement learning agents are trained to make these decisions autonomously and by adapting to the process environment. The project also investigates if, or how accurately, the improvement obtained via rescheduling can be predicted a priori. The methodology is applied to the rebalancing of an urban bike share system and the scheduling of mining operations.

2020

2023