Teaching

I teach a course “machine learning for energy systems” every fall semester at DTU. This is a mandatory course, featuring 5 ECTS and comprising 13 four-hour lectures, in the digital energy systems program. This course covers the applications such as wind power prediction, trading in the spot and balancing markets, control of a battery storage and large-scale generation scheduling. More details about course description, learning objectives, and the content are available below. 

46765: Machine learning for energy systems

Semester: Fall

Offered: 2023-2024

Lecture 1: Course introduction, Python introduction (Slides, Exercise and solution) 

Lecture 2: Machine learning introduction and overview of the course (Slides) 

Lecture 3: Scikit-learn introduction (Exercise and solution), Gurobi introduction (Exercise and solution)  

Lecture 4: Regression (Slides) 

Lecture 5: Wind trading in electricity markets (Slides) 

Lecture 6: Non-linear regression for wind prediction (Slides, Exercise and solution) 

Lecture 7: Lasso and Ridge regression (Slides, Exercise and solution), Validation and feature selection (Slides) 

Lecture 8: Clustering and classification (Slides, Exercise and solution) 

Lecture 9: Classification for optimal day-ahead scheduling (Slides) 

Lecture 10: Reinforcement learning for real-time control of an asset (part 1) (Slides) 

Lectures 11 and 12: Reinforcement learning for real-time control of an asset (part 2) (Slides, Exercise and solution) 

Lecture 13: Recap and potential future research ideas (Slides) 

Assignment 1

Assignment 2

Assignment 3