Big Data Framework for Analytics in Smart Grids and Applications on Electric Vehicle Loads

• Author / Creator

  Munshi, Amr

• The traditional electric grid based on centralized generation plants and unidirectional transmission and distribution systems is transitioning to a smart grid that is decentralized and multidirectional with high integration of information and communication technologies. With the rapid development of smart grids, large amounts of smart meters and sensors are deployed with huge coverage. As a result, large amounts of multi-sourced heterogeneous smart grid data are being produced. This massive amount of data needs to be sufficiently managed to increase the efficiency, reliability, and sustainability of the smart grid. Interestingly, the nature of smart grids can be considered as a big data challenge that requires advanced informatics techniques and cyber-infrastructure to deal with huge amounts of data and their analytics to take the smart grid a step forward in the big data era.
  In this thesis, a big data framework that potentially promotes innovative smart grid data analytics is presented. Further, the framework is developed to comply with the Lambda architecture that is capable of performing parallel batch and real-time operations on distributed data. Implementations of the frameworks on cloud-computing based platforms are presented, and various applications are applied on top of the framework, including visualization, load monitoring, and data mining. The framework is able to acquire, store, process and query massive amounts of smart grid data in near real-time, which is milliseconds in this study. This suggests that the framework is feasible in performing further smart grid data analytics.
  The second part of the thesis presents various smart grid applications that are applied on top of the smart grid big data framework. First, an unsupervised algorithm to extract electric vehicle charging loads (EVCLs) non-intrusively from the smart meter data is proposed. The proposed algorithm can run on low-frequency smart meter sampling data and requires only the real power smart meter measurement, which is the type of data recorded and communicated by most smart meters. Validation results on real aggregated residential household loads have shown that the proposed approach is efficient in extracting EVCLs and effective in mitigating the interference of other appliances, such as cloth dryers and air condition systems, that have similar load behaviors as electric vehicles (EVs). Secondly, a method to define flexibility for the collective EV charging demand is presented. This is achieved by analyzing the time-variable patterns of the aggregated EV charging behaviors. Furthermore, a case study on real residential data to analyze EV charging trends and quantify the flexibility achievable from the aggregated EV load in different time periods is presented. To verify the effectiveness of the approach, the EVCL extraction algorithm was applied on real residential datasets. The results of extracting the EVCLs from residential households were satisfactory. The extracted EVCLs were segmented into weekdays and weekends, and the flexibility achievable from the collective EV charging behavior was analyzed. Further, statistical indicators that represent time periods where trends in EV charging may occur are discussed. Finally, a method to group EV charging customers into clusters to reshape the aggregated EVCL is presented. This part of the thesis promotes the reliability and economical operation of smart grids. The utilized indicators based on statistical analysis can potentially assist operators and researchers in understanding time periods where trends in EV charging behaviors may arise and act accordingly.

• Subjects / Keywords

  • Electric vehicle
  • Smart grids
  • Big data

• Graduation date

  Fall 2019

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-e0c8-4217

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.