High Fidelity Modeling and Optimal Resilience Management in PEBB-based Power Distribution Systems
Abstract
Development of a power distribution corridor using power electronics building blocks (PEBB) renders unparallel flexibilities in agil,e reconfiguration and timely access to electric power in electric ships. PEBB, as the main building block, represents a cyber-physic,al system with embedded sensors that provide valuable information on the state of health and functionality. Within the context of th,e proposed power corridor, frequent reconfiguration of PEBBs, energy storage, and loads is foreseen. These transitions can lead to a, rich set of dynamic behaviors such as limit cycling, quasi-periodic, and chaos which usually lead to an increased magnitude of puls,ation on the dc-link voltage and potentially failure of the system.The primary challenges that may undermine the stability of power,corridor includes (a) vulnerable components such as capacitors and semiconductor switches that are subject to ageing and strong osci,llations, (b) wide-timescale control dynamics causing instability, and (c) the cyber layer (i.e. control and communication) being pr,one to cyberattacks and sensor faults which, in turn, can lead to compromised performance or even failure.As the placements of PEBBs, and operating conditions change, it becomes very difficult to capture the time-varying nature of the system in real-time. This call,s for a self-tuning model with coupled dynamics and time-varying parameters.Furthermore, access to high precision data is of high im,portance to development of such self-tuning model. However, polluted data due to measurement noise and malicious cyberattacks can un,dermine the fidelity of the incoming data. The unreliable data, in turn, can result in development of incorrect model. Moreover, in,certain conditions, system or components may become vulnerable , even under normal mo,e anticipated and prevented. It is, therefore, important to include a self-detection and anticipation/prevention algorithm.In additi,on, the multi-timescale, different switching frequencies, and presence of fast dynamics, especially during reconfiguration of the PE,BBs can lead to escalation of unwanted transients. Since system optimization and local PEBB controllers are designed separately, the, real time uncertainties of the local controllers can deviate from the setpoints during optimization intervals. To alleviate the abo,ve issues, it is imperative to develop a coordinated optimization and control methodology to maintain the desired performance and ov,erall reliability of the power corridor.To highlight the problems associated with the legacy control issues, a reduced scale version, of the power corridor will be designed and developed. The individual controllers used in this system employ PI as a point of refere,nce. This system will be used to verify the effectiveness of the proposed control architecture as compared to legacy controllers.The, objectives of the project are:(A) Development of a self-evolving model by integrating real-time data into physics-based model, to e,rithm using physical features of the system using deep neural network based hidden Markov model, physical model, and data to guarant,ee correct decisions in updating models, detecting cyberattacks, and to predict vulnerabilities.(C) Development of a reliability man,ench scale hardware of the power corridor for validation of the proposed concepts and demonstration purposes.Successful completion o,f this project will ensure high resiliency and stability for power corridor in the presence of intermittent changes in load/source,,cyberattacks, and coupled nonlinearity associated with the system. Approved for public release.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 13, 2022
- Source ID
- N000142212557
Entities
People
- Babak Fahimi
Organizations
- Office of Naval Research
- United States Navy
- University of Texas at Dallas