Reinforcement Learning based MIMO Controller for Virtual Inertia Control in Isolated Microgrids

Vjatseslav Skiparev, Juri Belikov, Eduard Petlenkov, Yoash Levron

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In this paper, we propose a multi-input multi-output controller for optimal control of nonlinear energy storage, using deep reinforcement learning (DRL) algorithm. This controller provides the frequency support in an isolated microgrid with high penetration of variable renewable energy sources and varying system inertia. To achieve an optimal control we redesigned neural network of actor and critic, simplified deep deterministic policy gradient (DDPG) rules, and reorganized the reward/punishment system. Simulation results show the efficiency of the proposed virtual inertia control architecture in several scenarios.

Original languageEnglish
Title of host publicationProceedings of 2022 IEEE PES Innovative Smart Grid Technologies Conference Europe, ISGT-Europe 2022
ISBN (Electronic)9781665480321
DOIs
StatePublished - 2022
Event2022 IEEE PES Innovative Smart Grid Technologies Conference Europe, ISGT-Europe 2022 - Novi Sad, Serbia
Duration: 10 Oct 202212 Oct 2022

Publication series

NameIEEE PES Innovative Smart Grid Technologies Conference Europe
Volume2022-October

Conference

Conference2022 IEEE PES Innovative Smart Grid Technologies Conference Europe, ISGT-Europe 2022
Country/TerritorySerbia
CityNovi Sad
Period10/10/2212/10/22

Keywords

  • Virtual inertia control
  • deep reinforcement learning
  • microgrids
  • renewable energy

All Science Journal Classification (ASJC) codes

  • Information Systems
  • Computer Networks and Communications

Fingerprint

Dive into the research topics of 'Reinforcement Learning based MIMO Controller for Virtual Inertia Control in Isolated Microgrids'. Together they form a unique fingerprint.

Cite this