General Description: 

This seminar focuses on the data-driven discovery of simulation models in industrial settings, providing a hands-on approach to understanding and optimizing production processes.

Students will start by designing and constructing production lines using Lego Spike. This activity will include developing comprehensive data-capturing pipelines to collect detailed raw event log data from their production lines. Next, students will explore advanced techniques for transforming raw event log data into simulation models, such as Petri nets. They will learn and apply data-driven model extraction methods, including process mining to discover workflow processes, statistical methods for fitting probability distributions and analyzing trends, and machine learning algorithms for modeling complex behaviors within the production process. Through these techniques, students will extract simulation models that reflect the real-world dynamics of their production lines. Finally, they will learn how to validate the extracted simulation models to ensure their accuracy and reliability.

By the end of the seminar, students will be equipped with the skills to build production lines, collect event log data, transform these data into actionable simulation models, and use these models to drive efficiency and innovation in industrial production settings.

Deliverables:

1 Report per team (20 pages, ACM Format) + Video of the finished physical line + Presentations (3) + Implementation files.

Registration:

Please briefly state your motivation (<= 200 words) for taking this course. Additionally, attach your Transcript of Records (Bachelors and Masters) and CV.

Grading relevant Parts:

Written Report (40%), Presentations (30%) and Implementation (30%).

Focus for presentations:

1. Introduction to the Project: Define the production line based on the provided requirements. Identify what data will be extracted and from where in the system. Outline the technologies and methodologies that will be used throughout the project. Prepare a project plan (e.g., a Gantt chart) and allocate tasks to each team member. 
2. Mid-project Presentation: Present the implemented LEGO production line, including a demonstration video and the generated event log data. Explain the methodology used for event log collection and process mining, and present the mined Petri net models derived from the event logs. Demonstrate the simulation of the extracted models and define the key performance indicators (KPIs) that will be used for model validation. 
3. Final Project Presentation: Deliver a comprehensive presentation of the completed project. Present the validation of the extracted simulation model against the defined KPIs and demonstrate the automatic updating of the Digital Twin in response to changes in the physical production line. Use the validated model to perform a what-if scenario analysis for system enhancement, including the implementation of the recommended improvement in the physical production system to evaluate whether the predicted benefits translate into real-world performance improvements. Conclude with future improvements and recommendations, and justify the task allocation by summarizing the contributions and effort of each team member.