Decentralized control and the analysis of electrical distribution systems
- This thesis studies the design and analysis of decentralized control over stochastic nonlinear systems. Different from the traditional centralized control, decentralized control considers multiple controllers, each with different information structure, actuating a system collaboratively. The thesis is composed of two parts. In the first part, we analyze the system stability of a distribution system with inverter-connected distributed power generation. Inverters are decentralized in nature. They are only allowed to communicate through local system measurements in voltage and current. No direct communication exists between these inverters. This convention of inverter control is nonlinear in the measured voltage and current. This work analyzes how the control policy of droop inverters affects the existing grid, and the robustness of the closed-loop system under disturbances. In the second part, we consider a theoretical problem in decentralized control over an adversarial network. We consider a networked system built on top of unreliable channels. These channels suffer from random information loss. The actuation signal of controllers are carried over a TCP-like protocol. In particular, we consider the sparsity information structure of controllers that are quadratically invariant. The problem is first formulated into a partially observable Markov decision process (POMDP). From the perspective of a fictitious player, we transform this decentralized problem into a centralized problem, which allows us to synthesize an optimal control policy with Kalman filters and value recursions.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Stanford University, Department of Electrical Engineering.
|Gorinevskiĭ, D. M
|Boyd, Stephen P
|Boyd, Stephen P
|Statement of responsibility
|Submitted to the Department of Electrical Engineering.
|Thesis (Ph.D.)--Stanford University, 2013.
- © 2013 by Chung-Ching Chang
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...