Section: Application Domains

Energy Management

Energy management, our prioritary application field, involves sequential decision making with:

• stochastic uncertainties (typically weather);

• both high scale combinatorial problems (as induced by nuclear power plants) and non-linear effects;

• high dimension (including hundreds of hydroelectric stocks);

• multiple time scales:

  • minutes (dispatching, ensuring the stability of the grid), essentially beyond the scope of our work, but introducing constraints for our time scales;

  • days (unit commitment, taking care of compromises between various power plants);

  • years, for evaluating marginal costs of long term stocks (typically hydroelectric stocks);

  • decades, for investments.

Significant challenges also include:

• spatial distribution of problems; due to capacity limits we can not consider a power grid like Europe + North Africa as a single “production = demand” constraint; with extra connections we can equilibrate excess production by renewables for remote areas, but not in an unlimited manner.

• other uncertainties, which might be modelized by adversarial or stochastic frameworks (e.g. techno- logical breakthroughs, decisions about ecological penalization).

We have had several related projects (Citines, a European (FP7) project; IOMCA, a ANR project), and we now work on the POST project, a ADEME BIA about investments in power systems. Our collaboration with company Artelys (working on optimization in general, and in particular on energy management) is formalized as an Inria ILAB.

 

Technical challenges: Our work focuses on the combination of reinforcement learning tools, with their anytime behavior and asymptotic guarantees, with existing fast approximate algorithms. Our goal is to extend the state of the art by taking into account non-linearities which are often neglected in power systems due to the huge computational cost. We study various modelling errors, such as bias due to finite samples, linearization, and propose corrections.

 

Related Activities:

• Joint team with Taiwan, namely the Indema associate team (see Section 9.4.1.1 ).

• Ilab METIS, in progress with Artelys (see Section 6.6 ) for industrialization of our work. In particular, the Crystal tool is adopted by the European Community (http://www.artelys.com/news/ 120/90/Energy-The-European-Commission-Chooses-Artelys-Crystal)

• Organization of various forums and meetings around Energy Management

• Visit of Edgar Galvan Lopez also includes applications to energy management, more precisely Demand-Side Management systems. In [40] , Differential Evolution is used to generate optimal plans to use the accumulators of electrical vehicles in order to reduce the peak household consumption loads.