Large systems such as manufacturing plants, hospitals, petrochemical refineries, airports, and amusement parks are characterized by processes, flows (material, patients, passengers, information), and capacity-constrained resources. Because of their complexity, they are often represented by simulation models (digital twins) so that their behaviour can be predicted under different design decisions. Exploring the design space of these systems requires both a simulation model and a systematic way to search the space and evaluate the results. Students will focus on these exploration techniques and use genetic algorithms to guide the search.
- Paraphrase a process for integrating system simulation with genetic algorithms to explore a multi-objective design space.
- Apply generic concepts of operations analysis (queueing and scheduling) to model a system’s operational performance in a particular domain.
- Apply data manipulation and visualization techniques to analyse system demand in the chosen domain.
- Interact with a client to modify the system an extension of the system simulation and design scenarios to address client-meaningful questions.
- Evaluate and present client-meaningful results of the demand analysis and design space exploration.
- Development and delivery of an oral presentation which summarizes time-based customer-arrival data, derive demand for critical resources and visualize resource-demand schedules distinguished by customer-type.
- Development and delivery of an oral presentation describing experiments to explore the trade-offs between competing objectives for a given digital twin.
- Demonstrate use of a genetic optimization library to identify Pareto-optimal design configurations.
- Propose and modify the digital twin in a meaningful way such as improved resource allocation.
- Define scenarios for evaluation to answer meaningful questions
- Development and delivery of an oral presentation presenting an executive summary of the design trade-offs discovered for the scenarios considered.