Loading...

Using Machine Learning Methods to Determine Extreme Weather-Related Outage Scenarios for Optimal Siting and Sizing of Distributed Generation and Energy Storage Facilities to Enhance Distribution Network Resiliencey

Mohammad Yari, Mohammad | 2021

272 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54284 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Hosseini, Hamid
  7. Abstract:
  8. Natural disasters such as storms, earthquakes, floods, etc. are among the factors that can damage the elements of the power grid and cause them to leave the grid. The result of these outages will be a widespread power outage, which will create many economic and social problems. Therefore, in recent years, the concept of resiliencey has been introduced in power system studies, which examines the extensive outputs in the network as a result of natural disasters.One way to improve the power system resiliency is to strengthen the network by upgrading existing equipment, installing new equipment, and modifying the operation of the network. Among these, the installation of distributed generation sources and energy storage facilities in appropriate locations is one of the most important proven solutions to improve the power system resiliency The main prerequisite for determining the location and size of distributed generation resources and energy storage facilities is to identify sensitive elements that are prone to network outages as a result of natural disasters. For this reason, if it is possible to predict which network elements are prone to go out of circuit during a natural disaster, precautionary measures can be taken to reduce the damage to the network, meaning that fewer subscribers will experience power outages. Among the methods used for this prediction are machine learning algorithms. Machine learning algorithms work by predicting the output of network elements using existing data and past system behavior. Which network elements are damaged and removed will lead to different scenarios. In this dissertation, after earthquake modeling, different natural disaster scenarios were determined by machine learning algorithms.After identifying the scenarios, network reconfiguration, installation of distributed generation resources and energy storage facilities were performed to improve network resiliencey. Also, for the first time, formulations were introduced to maintain the radiality so that the network is divided into several microgrids, each of which provides its own local loads independently of the other. Finally, based on the criterion of regret, the optimal plan was presented, which shows the installation location of distributed generation sources and energy storage facilities
  9. Keywords:
  10. Machine Learning ; Energy Storage System ; Distributed Generation ; Power System Resiliency ; Network Resiliance

 Digital Object List

 Bookmark

No TOC