Damien Ernst : Biobibliography & research career


Professor Damien Ernst receives the Blondel Medal

Portrait-ErnstDamien-V450

Academic background

2017 - today Ordinary Professor at the University of Liège. Affiliated with the Montefiore Research Unit.
2015 - 2016 Full Professor at the University of Liège. Holder of the EDF- Luminus Chair on Smart Grids, at the Montefiore Institute.
2011 - 2014 Lecturer at the University of Liège. Holder of the EDF-Luminus Chair on Smart Grids, at the Montefiore Institute.
2007 - 2011 Research associate of the FRS-FNRS. Affiliated with the "Systems and Modelling" Research Unit at the Montefiore Institute.
2006 - 2007 Teacher-researcher at SUPELEC (Rennes, France).
2003 - 2006 Research Fellow of the FRS-FNRS. Affiliated with the Stochastic Methods Department, Montefiore Institute.
2003 Graduated Doctor of Applied Sciences, University of Liège. With the greatest distinction. PhD under the supervision of Prof. Louis Wehenkel, Stochastic Methods Department.
1998 Graduated Civil Engineer Electrician and Mechanic, at the University of Liège. With the greatest distinction.

 

Visiting researcher stays in foreign laboratories

June 2004 - September 2004 Federal Institute of Technology Zürich (ETH, Switzerland).
March 2004 - May 2004 Massachusetts Institute of Technology (MIT, USA).
September 2003 - December 2003 Carnegie Mellon University (CMU, USA).
December 1999 Royal Institute of Technology in Stockholm (KTH, Sweden).
February 1999 - April 1999 Virginia Tech (VT, USA).

 

Research areas and scientific publications

Damien Ernst's scientific work has so far resulted in the publication of two books, twelve book chapters, more than fifty newspaper articles, and about one hundred and fifty articles in conference proceedings.

From its scientific production emerge five work groups that we present below. Most of this work was done in collaboration with other researchers and Damien Ernst's scientific contribution to this work was essential to its successful completion.

Transient (angular) stability of electrical systems

Chronologically, Damien Ernst's first research work is in the field of transient (angular) stability of electrical systems.

The techniques he helped to develop are based on reducing the electromechanical dynamics of a multi-machine system to that of a simplified model composed of a "critical machine" and an "infinite network". In this area, he has contributed in particular to the development of corrective control schemes aimed at mitigating the impacts of  loss of synchronism phenomena, for example through generation shedding. He has also worked on the development of techniques for the rapid identification of dangerous contingencies and the determination of preventive control actions to limit the risk of loss of synchronism.

Many researchers around the world have used this work to build their own analytical and control techniques for transient stability.

References

[1] "Model predictive control of HVDC power flow to improve transient stability". Y. Phulpin, J. Hazra and D. Ernst. In Proceedings of the Second IEEE International Conference on Smart Grid Communications (IEEE SmartGridComm), Brussels, Belgium, October 17-20, 2011.

[2] "A unified approach to transient stability contingency filtering, ranking and assessment". D. Ernst, D. Ruiz-Vega, M. Pavella, P. Hirsch and D. Sobajic. In IEEE Transactions on Power Systems, August 2001, Volume 16, No 3, pages 435-443.

[3] Transient Stability of Power Systems: A Unified Approach to Assessment and Control. M. Pavella, D. Ernst, and D. Ruiz-Vega. Kluwer Academic Publishers. Printed in the United States of America (2000). 255 pages. ISBN: 0-7923-8163-7.

Using reinforcement learning techniques  to control electrical systems

During his doctoral thesis [7], Damien Ernst studied reinforcement learning techniques to develop corrective control schemes to stem loss of synchronism phenomena. Reinforcement learning is an area of automatic learning aimed at developing intelligent agents capable of maximizing a reward signal from information collected during interactions with their environment. Reinforced learning is also often considered a Monte-Carlo approach to solving optimal control problems.

This first application of reinforcement learning in power grids has been followed by many others. For example, Damien Ernst has also applied learning by reinforcement to the damping of slow electromechanical oscillations [4,6], or to model the behaviour of different market players acting on an electrical system [5].

Damien Ernst's work has been pioneering in the use of these approaches from reinforcement learning in the context of electrical systems. Since then, many other researchers in this field have also made use of it.

References

[4] "Reinforcement learning versus model predictive control: a comparison on a power system problem". D Ernst, M Glavic, F Capitanescu, L Wehenkel. In IEEE Transactions on Systems, Man and Cybernetics. Part B, April 2009, Volume 2, pages 517-519.

[5] "A comparison of Nash equilibria analysis and agent-based modelling for power markets". T Krause, EV Beck, R Cherkaoui, A Germond, G Andersson, D Ernst. International Journal of Electrical Power & Energy Systems ; 2006, 28 (9), 599-607.

[6] "Power systems stability control: Reinforcement learning framework". D. Ernst, M. Glavic and L. Wehenkel. In IEEE Transactions on Power Systems, February 2004, Volume 19, pages 427-435.

[7] "Near Optimal Closed-Loop Control. Application to Electric Power Systems". D. Ernst. PhD thesis, University of Liège, Belgium, March 2003, 284 pages.

Learning by reinforcement and stochastic programming

Early in his research, Damien Ernst realized that existing methods in reinforcement learning and stochastic programming were not capable of dealing with the complexity of control and sequential decision problems encountered in many real applications. He also worked on the development of new algorithms for reinforcement learning [11,12] and stochastic programming [10], better adapted to the large scale problems in which he was interested.

In particular, he is the first author of the article "Tree-based batch mode reinforcement learning" published in 2005 [12], which has since established itself as one of the major references in the field of reinforcement learning. Many scientists in fields as diverse as medicine, finance and robotics have drawn inspiration from the algorithms proposed in this article - and more particularly from the "Fitted-Q Iteration" (FQI) algorithm - to build solutions to their own optimal sequential decision-making problems.

In recent years, Damien Ernst has supervised or supervised several doctoral theses in this extremely exciting field of research (e. g. [8,9]).

References

[8] "Benchmarking for Bayesian Reinforcement Learning". M. Castronovo, D. Ernst, A. Couetoux and R. Fonteneau. PLoS ONE 11(6), 2016, 25 pages.

[9] "Batch mode reinforcement learning based on the synthesis of artificial trajectories". R. Fonteneau, S.A. Murphy, L. Wehenkel and D. Ernst. Annals of Operations Research, Volume 208(1), September 2013, pages 383-416.

[10] "Multistage stochastic programming: A scenario tree based approach to planning under uncertainty". B. Defourny, D. Ernst and L. Wehenkel. In Decision Theory Models for Applications in Artificial Intelligence: Concepts and Solutions, 2011, Information Science Publishing, 51 pages.

[11] "Reinforcement Learning and Dynamic Programming using Function Approximators". L. Busoniu, R. Babuska, B. De Schutter and D. Ernst. Published by Taylor & Francis CRC Press, 2010. ISBN: 978-1-4398-2108-4.

[12] "Tree-based batch mode reinforcement learning". D. Ernst, P. Geurts and L. Wehenkel. In Journal of Machine Learning Research, April 2005, Volume 6, pages 503-556.

The electric super-grids

In recent years, Damien Ernst has been interested in how to develop and control super-electrical grids (continental or intercontinental), in particular to ensure better operation of very large electrical systems.

In particular, he worked on the problem of sharing the primary reserve between synchronous zones interconnected by direct current links. The scientific approach he has adopted to analyze and solve this problem is very original. Indeed, this is the first time that elements of the consensus theory, which has been extensively studied in recent years in control theory, have been used in this context to optimize the control of electricity networks [13].

Damien Ernst has also worked in recent years on the crystallization of a vision of the long-term evolution of the structure of electrical systems. This work led him to conclude that, thanks to the massive investments in renewable energy, a global super-grid linking most of the world's major centres of electricity consumption and production could emerge. This vision was published in 2013 in an article entitled "The Global Grid"[14]. Currently, several researchers from Damien Ernst's team are working on this concept of a "Global Grid", focusing initially on the construction of an electrical mega-connection between Greenland and France [15], a plausible step to then electrically connect the European continent with the American continent.

Many other researchers in the field of electricity networks, as well as some international companies  (notably in Europe and China) are interested in these subjects.

References

[13] "Cooperative frequency control with a multi-terminal high-voltage DC network". J. Dai, Y. Phulpin, A. Sarlette and D. Ernst. Automatica, Volume 48, Issue 12, December 2012, pages 3128–3134.

[14] "The global grid". S. Chatzivasileiadis, D. Ernst and G. Andersson. Renewable Energy, Volume 57, September 2013, pages 372–383.

[15] "Harvesting Greenland’s katabatic winds for European electricity supply". D. Ernst et al. To be submitted, non disponible on-line mais joint comme annexe au dossier de candidature.

Integration of renewable energies into distribution networks

Since 2011, Damien Ernst has held the Smart Grids Chair at the University of Liège, and works very closely with various actors in the Belgian electricity sector to effectively integrate renewable energies into the electricity distribution networks. The scientific approach he has developed is quite original in the sense that he is working on an integrated development of all the elements of the decision-making chain (interaction models between stakeholders, investment strategies, forward planning, real-time control) necessary to solve this problem in a viable way.

Damien Ernst's work in this context has led to leading scientific publications [16,17,18] and already to the implementation of some of his research results on distribution networks in Belgium. This work also led to the creation of BLACKLIGHT ANALYTICS (Damien Ernst is one of its co-founders) which markets software solutions based on this work.

References

[16] "Active network management for electrical distribution systems: problem formulation, benchmark, and approximate solution". Q. Gemine, D. Ernst and B. Cornélusse. In Optimization and Engineering, September 2017, Volume 18(3), September 2017, pages 587–629

[17] "DSIMA: A testbed for the quantitative analysis of interaction models within distribution networks". M. Sébastien, Q. Louveaux, D. Ernst and B. Cornélusse. In Sustainable Energy, Grids and Networks, Volume 5, 2016, pages 78-93.

[18] "Active management of low-voltage networks for mitigating overvoltages due to Photovoltaic Units". F. Olivier, P. Aristidou, D. Ernst and T. Van Cutsem. In IEEE Transactions on Smart Grid, Volume 2(7), March 2016, pages 926-936.

Links with industry and electrical operators

Most of Damien Ernst's work has resulted in scientific publications which, in most cases, propose solutions to very concrete problems encountered in our society and, in particular, in the field of electrical systems. Many other researchers around the world have used his work as a basis for developing solutions to similar problems and which have also led to concrete applications.

At the beginning of his post-doctorate, Damien Ernst stayed in the control centre of the Greek transmission system operator to put into service the analysis and control module for transient stability developed at the University of Liège.

Currently, he leads a team of researchers whose objective is to provide electricity distributors with decision-making software to effectively integrate renewable energies and flexible loads into their grid. In 2017, he also co-founded BLACKLIGHT ANALYTICS to industrialize these software products. To date, this company markets two main products: a software to optimize the capacity of renewable energy that can be connected to distribution networks and solutions to intelligently modulate the production of renewable energy sources in order to avoid congestion and tension problems in distribution networks. The main feature of the software developed in this context is the extremely high performance of the decision-making algorithms, built thanks to the skills of  his research team not only in the field of electrical networks, but also in optimization and artificial intelligence.

It should also be noted that for the past two years, Damien Ernst has been working closely with Prof. Bertrand Cornélusse of the University of Liège (Chair of Micro-Networks) and with RESA (local distribution network manager), Nethys and CMI on the development and industrialization of an "Energy Management System" for micro-networks based on artificial intelligence techniques [1].

References

[1] “Efficient management of a connected microgrid in Belgium”. B. Cornélusse, D. Ernst, L. Warichet and W. Legros. Proceedings of the 24th International Conference on Electricity Distribution, June 2017, 4 pages.

Contact

Pr Damien ERNST

Université de Liège, Montefiore Institute, Bâtiment B28

Quartier Polytech

Allée de la découverte, 10

4000 Liège

BELGIQUE

 

Portable : +32 493 04 25 68

E-mail : dernst@uliege.be

Website : www.damien-ernst.be

Facebook : www.facebook.com/damien.ernst.5

Twitter @DamienERNST1

 

Share this page