*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Thesis Title:
Decentralized Optimization and Analytics for Large Scale Power System Problems
Advisors:
Dr. Nagi Gebraeel, School of Industrial and Systems Engineering,
Georgia Tech
Dr. Edmond Chow, School of Computational Science and Engineering,
Georgia Tech
Committee members:
Dr. Santanu Dey, School of Industrial and Systems Engineering, Georgia
Tech
Dr. Andy Sun, School of Industrial and Systems Engineering, Georgia
Tech
Dr. Michael Rabbat, Facebook AI Research, Montreal
Date: Thursday, October 8th, 2020
Time: 830am - 10am, EST
Meeting URL (for BlueJeans):https://bluejeans.com/520087903
Meeting ID (for BlueJeans):520
087 903
Abstract:
Large scale power networks form the backbone of the global energy infrastructure. Power system optimization problems aim at better utilization
of system resources and form a significant part of power systems research. However, large scale planning and optimization problems demand efficient computational schemes that respect the data privacy of asset owners and operators as well. As a result, decentralized
computational paradigms for large scale planning problems in power systems are gaining popularity.
In this thesis, we explore novel ways to solve computationally challenging planning and analytics problems in a decentralized manner using
synchronous as well as asynchronous computational models. We specifically focus on decentralized formulations of unit commitment, joint operations and maintenance, differential privacy based unit commitment and maintenance as well as a blockchain based, decentralized
analytics methodology for replay attack detection.
In Chapter 2, we establish a synchronous solution mechanism to the decentralized sensor driven optimization problem that provides a comparable
solution to the centralized method. In Chapter 3 we tackle the more fundamental decentralized UC (DUC) problem and explore an asynchronous solution for its computational benefit. We show that the asynchronous approach presented in Chapter 3 potentially leads
to faster solution times showing good computational promise.
In Chapter 4, we strengthen our solution to the DUC problem with the help of privacy preserving valid inequalities that are computed asynchronously
leading to a significant improvement in solution quality. We introduce an interleaved binary approach to our asynchronous method that improves convergence times for the asynchronous solution drastically.
In Chapter 5, we present our work that is driven by a subgradient based solution to the integrated sensor driven, decentralized maintenance
and UC problem for planning horizons spanning many months. In this approach, by dualizing the maintenance limit constraint that is coupled across weeks, we parallelize weekly operations locally with the help of the subgradient method. We orchestrate multithreading
for every subproblem while employing the distributed memory paradigm to communicate among regions.
In Chapter 6 we propose a differential privacy driven approach geared towards decentralized formulations of mixed integer operations and maintenance
optimization problems that protects network flow estimates. We prove strong privacy guarantees by leveraging the linear relationship between the phase angles and the flow. To address the challenges associated with the mixed integer and dynamic nature of the
problem, we introduce an exponential moving average based consensus mechanism to enhance convergence, coupled with a control chart based convergence criteria to improve stability.
In Chapter 7, we propose a blockchain based decentralized framework for detecting coordinated replay attacks with full data privacy. We develop
a Bayesian inference mechanism employing locally reported attack probabilities that is tailor made for a blockchain framework. We compare our framework to a traditional decentralized algorithm based on the broadcast gossip framework both theoretically as well
as empirically. With the help of experiments on a private Ethereum blockchain, we show that our approach achieves good detection quality and significantly outperforms gossip driven approaches in terms of accuracy, timeliness and scalability
