Knowledge Graph Representation of Power System and Data-Driven Analysis of Outages

• Author / Creator

  Kor, Yashar

• Weather-related power outages in the distribution grid have a significant impact on the grid reliability - they impose a high cost on power utilities and considerable inconvenience to customers. Improvements in monitoring and data collection practices, as well as advanced data processing methods, provide opportunities for comprehensive modeling and analysis of grid operations. At the same time, they allow for better understanding and handling of reasons for degradation of service quality due to power outages, weather patterns, and asset-related performance.

  The thesis focuses on applying Machine Learning and Computational Intelligence methods for the analysis and processing of power distribution system data. We design and develop a collection of data-driven algorithms, methods, and procedures for the investigation of relations between power grid, outage, and weather data. Additionally, they constitute a framework suitable for building a comprehensive system for analyzing and predicting weather-related outages and their severity. We propose Weather outage Prediction System (WoutPS) for forecasting outages based on multiple data-driven outage prediction models combined with a reasoning framework based on Dempster-Shafer theory (DST), as well as Knowledge Graph-based representation of distribution grid topology (GridKG) suitable for integration of data characterizing different aspects of the distribution system.

  Three different architectures of a system for predicting types of weather-related outages are proposed and evaluated. Weather and outage data are utilized for model development and evaluation of their performances. The developed system is capable of identifying the probability of outage occurrences with a focus on identifying outages caused by extreme wind, wet snow, and icing. An analysis of the prediction results is provided.

  The thesis includes details of developing a novel knowledge graph-based representation of a distribution system. The graph, called GridKG, integrates a variety of data: system topology, information about its components and customers, as well as data collected during systems events, in particular, power system outages. As a result, a comprehensive representation of a distribution system is obtained. We anticipate that the proposed way of representing a power distribution grid will lead to the discovery of novel ways of augmenting and predicting its reliability. We show the benefits of such representation: evaluation the impact of power outages on consumers in the power system without and with Distributed Energy Resources.

• Subjects / Keywords

  • knowledge graph
  • data-driven analysis
  • power outage
  • predictive model

• Graduation date

  Fall 2021

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-3svv-bd20

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.