MDP based real time restoration for earthquake damaged active distribution systems

Arpali O. Y., Yilmaz U. C., GÜLDÜR ERKAL B., AYDIN GÖL E., GÖL M.

Electric Power Systems Research, vol.218, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 218
  • Publication Date: 2023
  • Doi Number: 10.1016/j.epsr.2023.109230
  • Journal Name: Electric Power Systems Research
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, PASCAL, Environment Index, INSPEC
  • Keywords: Decision support, Disaster management, Distribution systems, Markov Decision Process, State estimation
  • Hacettepe University Affiliated: Yes


After a disaster, presence of electricity becomes even more crucial compared to its role in daily life. In this paper, an online decision support method is developed to restore medium voltage active distribution systems after an earthquake. The proposed method employs Markov Decision Process (MDP) to determine the sequence of actions which minimizes the expected restoration time. The method firstly predicts the health condition of the system components based on the Probability of Failure (Pf). These Pf values were priorly computed and plotted against Peak Ground Acceleration (PGA) values. Then a sequence of restoration actions, i.e. the restoration strategy, for the system operator is determined by considering the prediction of health conditions and power flow analysis results. During the field operation, if an unexpected situation, i.e. unexpected state of a component or a significant variation in electrical quantities, is encountered by the system operator, the proposed method updates the restoration strategy by considering those variations from the initial prediction.