Abstract
In event-driven rescheduling, new re-optimization procedures are triggered when obtaining new information that indicates the current schedule to be outdated. Critical design aspects of such an algorithm are the definition of the trigger event and the allocated computing time for a new rescheduling procedure. We treat both of these design aspects as continuous control parameters. Nevertheless, finding the best-suited control parameter combination for a given operating environment may be computationally expensive, as it requires simulating the process with many candidate combinations. We use surrogate-based optimization to reduce the computing cost of optimizing the control parameters. We demonstrate the method on real-time rebalancing of a bike sharing system and investigate the sensitivity of the optimized parameters to changes in the operating environment.
| Original language | English |
|---|---|
| Title of host publication | Computer Aided Chemical Engineering |
| Editors | Yoshiyuki Yamashita, Manabu Kano |
| Number of pages | 6 |
| Publisher | Elsevier |
| Publication date | Jan 2022 |
| Pages | 1255-1260 |
| ISBN (Electronic) | 9780323851596 |
| DOIs | |
| Publication status | Published - Jan 2022 |
| MoE publication type | A3 Book chapter |
Publication series
| Name | Computer Aided Chemical Engineering |
|---|---|
| Volume | 49 |
| ISSN (Print) | 1570-7946 |
Bibliographical note
Funding Information:Financial support from the Academy of Finland is gratefully acknowledged (project RELOOP, decision number 330388). The calculations presented in this work were performed using computer resources within the Aalto University, School of Science, “Science-IT” project.
Publisher Copyright:
© 2022 Elsevier B.V.
Fields of Science
- bike sharing rebalancing
- event-driven rescheduling
- logistics
- optimization
- surrogate modelling
- 113 Computer and information sciences