B9133 - DIGITAL TWIN

Learning outcomes

At the end of the course, the student will be able to build a Digital Twin of real-world entities and understand the processes of data exchange between the real entity and its digital twin, as well as the implications on their behaviors. During the course,students will become familiar with the basics of data analysis using mathematical models and machine learning. They will also gain confidence with methods for acquiring and processing data from complex real-world scenarios through methodologies such as Mobile Crowdsensing and Federated Learning. The students will learn to apply digital models of objects, processes, or real-world systems, synergistically combining simulation-based approaches with data-driven approaches. The contemplated applications range from monitoring (dashboards and rendering) to modeling tools that support (decision support, design support, what-if analysis) the design, operation, and management of complex systems. The course will cover technical, technological, and practical aspects, highlighting the use of digital twins in various domains such as monitoring, resource management, production in engineering and industry, and smart cities.

Course contents

• Introduction to Digital Twins and multidisciplinarity
• The IoT data pipeline
• Physics-based modeling
• Discrete-event simulation
• Data assimilation and uncertainty
• Time series analytics
• Physics-informed Machine Learning
• Anomaly detection
• Decision processes
• Reinforcement Learning

Readings/Bibliography

• Walid M. Taha, Abd-Elhamid M. Taha, Johan Thunberg, Cyber-Physical Systems: A Model-Based Approach.
• Steven L. Brunton, J. Nathan Kutz, Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control.
• Fundamental scientific papers online.

Teaching methods

The course takes place in the second semester and is structured as in-person lectures held in Italian, with course materials provided in English.

In particular, after presenting introductory and architectural notions, lectures will build up from the mathematical foundations of physical system modeling, towards more advanced data-driven techniques for both modeling and control. The presented concepts will be accompanied by short demos using python carried out live in class by the instructor. This approach fosters understanding of the principles and their practical implementation, enabling students to independently develop both the fundamentals and being able to apply them into a real scenario.

Given the type of activities and teaching methods adopted, attendance in this course requires prior completion by all students of Modules 1 and 2 of the safety training for study environments, available in e-learning mode at: https://elearning-sicurezza.unibo.it/

Assessment methods

The course requires the completion of a project (individual or in groups of two) involving the development of a Digital Twin. Students will need to propose their own use case coming from any domain (manufacturing, agriculture, smart cities) and use open data, certified simulators, or sensors in a real system as a ground truth. The project must then be discussed during an oral exam session, where the students will need to justify their choices and report their results. Before starting the development, the project must be agreed upon in advance with the instructor, preferably by sending an email with an extensive description.

By the deadline, students will need to deliver the code (possibly a link to a versioning system, like GitHub), alongside a written report in electronic format (PDF) that includes a description of how the application works supported by screenshots, design choices, and all references to materials, developed code, and documentation sources.

There are normally six annual submission deadlines: June, July, September, November, January, and February. Projects must be submitted by the indicated deadlines (communicated via the course mailing list and on the VIRTUALE course portal), exclusively through the course’s VIRTUALE platform, by uploading the report and a link to the implementation. After submission, students are usually invited to the oral exam within one or two weeks following the deadline. The exam date is scheduled in advance on Almaesami and published on the course newsgroup to allow interested students to attend. The exam takes place in person and will consist in a discussion of the project's design choices alongside questions about related topics in the course, to assess the student's preparation. The instructor determines if the discussion is sufficient, and proposes a grade. If it reaches or exceeds 30/30, honors (lode) may be awarded.

Students with Specific Learning Disorders (SLD) or temporary or permanent disabilities: students are encouraged to contact the University's dedicated support office as early as possible (https://site.unibo.it/studenti-con-disabilita-e-dsa/it ). The office will propose any appropriate accommodations for eligible students. Such accommodations must be submitted to the course instructor for approval at least 15 days in advance. The instructor will assess their suitability in relation to the intended learning outcomes of the course.

Regarding assessment, a limited, transparent, and non-substantive use of AI is permitted for support activities, such as summarization and rephrasing. The substantial use of AI to complete any part of the assessment is not permitted.

Teaching tools

Slides, personal computer, projector, online references.

Office hours

See the website of Federico Montori