Improved solver performance on scheduling problems with start-up time
The research consists in creating a scheduling tool which aims to schedule a series of tasks on a production line composed of a single machine in the food industry. The planning horizon consists of weeks 4 to 8.
The research consists in creating a scheduling tool which aims to schedule a series of tasks (jobs) on a production line composed of a single machine in the food industry. The planning horizon consists of weeks 4-8. The start-up time between tasks should be considered and minimized in order to improve the performance of the production line.
The research project is being conducted in collaboration with the company Biscuits Leclerc.
The company shared some of their data in order to serve as instances. The scheduling includes constraints specific to their reality such as a limit of tasks that can be performed during one single day or the requirement that certain tasks must begin at the beginning of the week. To achieve this, we use constraint programming, an optimization technique derived from artificial intelligence.
The research project involves the following steps:
- the creation of a model representing the reality of the business;
- a branching heuristic adapted to the problem;
- the use of a meta-heuristic aimed at improving the solution;
- and the implementation of the tool in the company.
IID Principal Investigators of this project
Jonathan Gaudreault Research Thrust 1 – Physical Environment
Université Laval Director, CRISI (Research Consortium for Industry 4.0 Systems Engineering)
Faculty of Science and Engineering
Department of Computer Science and Software Engineering
Director, CRISI (Research Consortium for Industry 4.0 Systems Engineering)
Industry 4.0 Internet of Things Logistics Machine Learning Operations Research
Claude-Guy Quimper, Jonathan Gaudreault, Nicolas Blais (Université Laval)
Project Funding: 2018-2021
Let’s keep in touch!
Would you like to be informed about IID news and activities? Subscribe now to our monthly newsletter.