Computing an Optimal Control Policy for an Energy Storage

Abstract : We introduce StoDynProg, a small library created to solve Optimal Control problems arising in the management of Renewable Power Sources, in particular when coupled with an Energy Storage System. The library implements generic Stochastic Dynamic Programming (SDP) numerical methods which can solve a large class of Dynamic Optimization problems. We demonstrate the library capabilities with a prototype problem: smoothing the power of an Ocean Wave Energy Converter. First we use time series analysis to derive a stochastic Markovian model of this system since it is required by Dynamic Programming. Then, we briefly describe the "policy iteration" algorithm we have implemented and the numerical tools being used. We show how the API design of the library is generic enough to address Dynamic Optimization problems outside the field of Energy Management. Finally, we solve the power smoothing problem and compare the optimal control with a simpler heuristic control.
Type de document :
Communication dans un congrès
Pierre de Buyl and Nelle Varoquaux. 6th European Conference on Python in Science (EuroSciPy 2013), Brussels, Belgium, Aug 2013, Brussels, Belgium. pp.51-58, 2014
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-00988072
Contributeur : Pierre Haessig <>
Soumis le : mercredi 7 mai 2014 - 11:42:04
Dernière modification le : mercredi 5 juillet 2017 - 01:01:38

Identifiants

  • HAL Id : hal-00988072, version 1
  • ARXIV : 1404.6389

Collections

Citation

Pierre Haessig, Thibaut Kovaltchouk, Bernard Multon, Hamid Ben Ahmed, Stéphane Lascaud. Computing an Optimal Control Policy for an Energy Storage. Pierre de Buyl and Nelle Varoquaux. 6th European Conference on Python in Science (EuroSciPy 2013), Brussels, Belgium, Aug 2013, Brussels, Belgium. pp.51-58, 2014. <hal-00988072>

Partager

Métriques

Consultations de la notice

131