Johnn, Syu-Ning; Charitopoulos, Vassilis M; (2024) A Reinforcement Learning Framework for Online Batch Process Scheduling. In: Manenti, Flavio and Reklaitis, Gintaras V, (eds.) Computer Aided Chemical Engineering. (pp. 1783-1788). Elsevier: Amsterdam, The Netherlands.

Text 436-johnn.pdf - Accepted Version Access restricted to UCL open access staff Download (405kB)

Abstract

Optimisation-based batch scheduling methods serve as a practical response strategy in the process industries as a coordination of planning and execution. The efficiency and adaptability of the method are of paramount importance, especially when dealing with frequent modifications to existing decisions caused by uncertainties and unforeseen realworld disturbances, with goal of achieving substantial financial profitability. Reinforcement Learning, compared to many classic techniques, has the advantages of learning from existed experiment and generalise to unknown scenarios, and thus automating the process with higher flexibility and adaptability. In this work, we propose a RL-based method by transferring a batch process scheduling problem into a Markov Decision Process framework and train the agent to learn to build up task sequences to optimise the production schedule. The results show that our method achieves good computational efficiency and adaptability.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item