Predictive models of Renewable electricity generation and electricity consumption profiles through Fourier Transform analysis

Master Thesis or Semester Project
Energy System Modeling
Renewable predictive model
Signal processing
Machine Learning
Grid stabilisation
Storage technologies
Sustainable Energy Transition

Nexi

Predictive models of Renewable electricity generation and electricity consumption profiles through fourier transform analysis

Semester/Master project

Fall 2026

Context

Today’s energy supply still heavily relies on fossil fuels, releasing greenhouse gases and damaging ecosystems. The society is moving to renewable alternatives like solar and wind energy, which are promising for decarbonized electricity production. Yet, they require balancing intermittency through clever complementarization strategies and/or storage technologies such as Battery systems or power-to-X-to-power devices, where X represents storable molecules for future power generation. In this context, developping a location-based predictive model for RE electricity production and consumption profiles can be very valuable for understanding the potential of these technologies. In this regard, Fourier representations of these time-varying signals can provide promising insights through the key frequency components. Indeed such a representation of these highly intermittent, yet periodic, signals can enable:

  1. Predictive modelling for both production and consumption based on geographical, economical and societal features.
  2. Defining complementary combinations of RE-energy production profiles
  3. Quantifying remaining and unavoidable curtailement/storage requirements.
  4. Dimensionning storage technologies based on their intrinsic characteristics such as (dis-)charging efficiencies and self-discharge rates (Battery: short-term, rSOC: long-term)

This framework would provide a fair and data-driven comparison of storage technologies and their potential around the world based on few key parameters only.

The Tasks

The project consists of the following main steps:

  • Understand the developed methodology for RE-electricity generation profiles.
  • Understand and improve the developed methodology for electricity consumption profiles.
  • Quantify the curtailement/storage needs on a worlwide basis.
  • Consider different storage technologies and handle multi-period optimization.
  • Create a Streamlit-app to generate visual and interpretable representations of the key results.

Skills

  • Basics of signal processing

  • Python programming

  • General knowledge on energy system modelling and Optimization

  • Interest in storage technologies.

  • Lectures:

    • Energy Conversion and Renewable Energy
    • Applied Data Analysis (recommended, not required)

Practical information

This study can consist of a 2-steps approach, starting with a semester project and following up with a Master thesis (if interest and dedication). It can also directly be a Master project.

The project will be conducted in IPESE lab in Sion (EPFL Valais) at about 1h05 from Lausanne train station. Possibility of meeting on the campus in Lausanne (mainly Mondays) once the project is up and running.

Any transportation fees between Lausanne and Sion will be reimbursed by the lab. English is required for the end report, but the supervision can be carried out in french, english or german.

Supervisor

The project will be supervised by Arthur Waeber. If interested, please contact me via e-mail attaching your CV and a short motivation letter.

mailto: arthur.waeber@epfl.ch;