Grant: Alliance Grants
École Polytechnique de Montréal
$65,700.00
Machine Learning-accelerated Power Electronics Modelling for Real-Time Simulations
FY 2024-25
Summary
Power electronics converters are essential for integrating renewable power generation into electric grids. As their number is growing, it becomes critical to carefully assess their potential impacts on power systems, e.g., with regard to stability, reliability, and safety. Characterizing the impacts of power electronics converters - and ultimately mitigating these impacts - entails running computational resources- and time-intensive time-domain offline and real-time simulations of the converters and the power systems. Power electronics converters are intricate devices that require involved physical modelling, e.g., to capture their switching behaviour or their highly nonlinear response. This leads to lengthy simulations which in turn implies that many model simplifications are required to reach the fast timescale needed for real-time simulations. In this research program, we propose new machine learning- and optimization-based methods for the fast, reliable, and trustworthy real-time simulation of power electronics converters. Leveraging machine learning will allow us to circumvent the modelling burden of power electronics. Our research program is devised into three main objectives: 1) build comprehensive learning-based models for power electronics simulation, 2) design the computational engine to support combined physics- and learning-based simulation on heterogeneous hardware like central processing units, graphical processing units, and field programmable gate arrays, and 3) robustly identify the safe operating space of a power electronics converter using a minimum number of time-domain simulations. The results will be thoroughly validated through numerical simulations and hardware-in-the-loop experimentation. Our research will yield fast numerical analysis tools for power systems without compromising on simulation accuracy. Our work will result in efficient digital replicas of power electronics converters-dominated grids. These digital models will enable in-depth stability and reliability studies, thus facilitating the integration of renewables into Canadian grids.
Program
Alliance Grants
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
593314-2024
Name nom
Lesage-Landry, Antoine A
Department département
Génie électrique
Organization id
64
Institution établissement
École Polytechnique de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2024
Award amount
65700
Program id
ALLRP
Program name en
Alliance Grants
Program name fr
Subventions Alliance
Group en
RESEARCH PARTNERSHIPS
Group fr
PARTENARIATS DE RECHERCHE
Committee code
96
Committee name en
RPP Internal Decision Cttee
Committee name fr
Cté de décision interne - PPR
Area of application code
304
Area of application group en
Energy resources
Area of application group fr
Ressources d'énergie
Area of application en
Electrical energy
Area of application fr
Énergie électrique
Research subject code
2517
Research subject group en
Electrical and electronic engineering
Research subject group fr
Génie électrique et électronique
Research subject en
Industrial and power electronics
Research subject fr
Électronique de puissance et électronique industrielle
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Machine Learning-accelerated Power Electronics Modelling for Real-Time Simulations
Keywords
grid forming inverters; inverter-based resources; machine learning; neural network; optimization; power electronic converters; power system stability; real-time simulations; renewable power systems; robust optimization
Application summary
Power electronics converters are essential for integrating renewable power generation into electric grids. As their number is growing, it becomes critical to carefully assess their potential impacts on power systems, e.g., with regard to stability, reliability, and safety. Characterizing the impacts of power electronics converters - and ultimately mitigating these impacts - entails running computational resources- and time-intensive time-domain offline and real-time simulations of the converters and the power systems. Power electronics converters are intricate devices that require involved physical modelling, e.g., to capture their switching behaviour or their highly nonlinear response. This leads to lengthy simulations which in turn implies that many model simplifications are required to reach the fast timescale needed for real-time simulations. In this research program, we propose new machine learning- and optimization-based methods for the fast, reliable, and trustworthy real-time simulation of power electronics converters. Leveraging machine learning will allow us to circumvent the modelling burden of power electronics. Our research program is devised into three main objectives: 1) build comprehensive learning-based models for power electronics simulation, 2) design the computational engine to support combined physics- and learning-based simulation on heterogeneous hardware like central processing units, graphical processing units, and field programmable gate arrays, and 3) robustly identify the safe operating space of a power electronics converter using a minimum number of time-domain simulations. The results will be thoroughly validated through numerical simulations and hardware-in-the-loop experimentation. Our research will yield fast numerical analysis tools for power systems without compromising on simulation accuracy. Our work will result in efficient digital replicas of power electronics converters-dominated grids. These digital models will enable in-depth stability and reliability studies, thus facilitating the integration of renewables into Canadian grids.