Starting Researcher Position: AI-driven projections of future wind power Full-time

Contexte et atouts du poste

Context

This Starting Research Position (SRP) will be co-supervised by Claire Monteleoni at INRIA Paris, and Boutheina Oueslati, Emmanuel Neau, and Yannig Goude, at EDF Lab, Saclay. The position will by financed by an INRIA-EDF Défi and a Choose France Chair in AI.

Mission confiée

Research project

Overview

This INRIA Starting Research Position will focus on improving projections and reducing uncertainty on future wind trends and anticipating periods of low wind, using innovative approaches based on new machine/deep learning techniques.

Transitioning our energy preduction to renewable sources is a clear path towards reducing CO2 emissions, which in turn is key to mitigating the most severe risks of of climate change. Most of these sources (e.g., solar, wind, and hydro) are variable in the short term, depending on local weather conditions, but are also evolving in a changing climate. The latest IPCC report suggests a decrease in average wind in Europe of between 8 and 10% for a warming scenario of 1.5°C and an increase in strong winds. Significant uncertainties are, however, still present on the processes that drive this future decrease in wind, in particular on the respective contributions of climate change and internal variability (Carvalho et al., 2021, Wohland et al., 2021). Therefore, current wind turbines may not be well-adapted to future wind conditions, either in their location or in their operating characteristics. Understanding and anticipating the future evolution of wind and its impact on wind power production is an important issue for the current and future electricity system in order to ensure its proper functioning in terms of flexibility but also means of adaptation.

Approach

Building on past work by members of the hosting teams on statistical and machine learning-based methods for downscaling spatiotemporal data, this project will explore existing methods and develop new methods to improve wind power forecasting at the necessary spatial and temporal scales. These AI-driven approaches will be used to analyse the evolution of the historical wind in Europe, the associated atmospheric dynamics and the link with the evolution of roughness based on observations, in order to reduce uncertainty on future predictions.

Specific tasks include:

• Develop and implement vertical wind extrapolation and spatio-temporal downscaling methods based on machine/deep learning methods
• Advance the understanding of the future evolution of wind production under the effects of climate change and also of the transformation of the wind farm (new installations and more efficient technologies) by addressing the following questions.
• Are climate models able to reproduce the observed trends of the wind as well as the associated atmospheric dynamics in a historical climate?

• How will these trends evolve under different future climate scenarios and what are the associated uncertainties?

• What will be the future wind production in Europe with a massive integration of offshore wind power?

Principales activités

1. Review existing literature on projecting future wind power generation
2. Design and implement pipelines that include machine learning for this application
3. Design and conduct empirical studies using the learned models for forecasting wind power generation, and comparison to state-of-the-art
4. Prepare, submit for publication, present, and disseminate research findings
5. Serve as a research mentor for doctoral students on related projects

Compétences

Technical skills and level required :

• Candidate must hold a PhD in Computer Science (Informatique), statistics, mathematics, or related fields
• Machine learning, data mining, statistics, and/or AI coursework and/or projects
• Familiarity with modern machine learning / deep learning software, tools, pipelines

Languages :

• Written competency in English
• Oral competency in English or French

Avantages

• Subsidized meals
• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Teleworking and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Social security coverage