Toward Distributed Control for Autonomous Electrical Energy Systems (AEESs)

Tuesday, September 10, 2019 - 3:00pm to Wednesday, September 11, 2019 - 3:55pm

Event Calendar Category

LIDS Thesis Defense

Speaker Name

Xia Miao

Affiliation

LIDS

Building and Room Number

32-D677

Abstract

THESIS COMMITTEE:
Dr. Marija Ilic (Thesis Supervisor)
Prof. Alexandre Megretski
Prof. James L. Kirtley Jr.
 
ABSTRACT:
In this thesis, we study the problem of enabling autonomous electrical energy systems (AEESs) by means of distributed control. Well-known concepts from dynamical systems are utilized by introducing a novel modeling of electrical energy systems and by further imposing additional quality of service (QoS) constraints observed in the EESs.
 
The proposed approach consists of five parts:

(1) We propose a modular modeling approach that represents a general EES as a negative feedback configuration comprising a planar electrical network subsystem HN whose components are two-port network elements; and a subsystem HS whose components are single-port elements, such as controllable power sources and uncontrolled power loads. Input-output modeling of each component is in terms of power and voltage, respectively. This is motivated by the basic functionality of balancing power supply and demand at the acceptable QoS measured in terms of frequency and voltage deviations from the nominal AC waveforms.
 
(2) We propose modular specifications for components of HS and HN so that these system functionalities can be achieved.  For the feasibility requirements, we require each stand-alone component to be BIBO. These feasibility conditions are given in terms of input, output and state initial conditions assuming disturbances caused by uncontrolled loads and control saturation are known and bounded. For the stability requirements, incremental passivity conditions are proposed by defining input, output and storage function as instantaneous power deviation, voltage deviation and incremental stored energy for each component of subsystem HN, and, voltage deviation, instantaneous power deviation and incremental stored energy for each component of subsystem HS, respectively.
 
(3) We propose modular distributed control of controllable components in HS and HN so that modular feasibility and stability conditions are met. For controllable components in HS, feedback linearizing control (FBLC) is designed so that the component is incrementally passive and finite gain stable. The same control principle is shown to be effective for electrical machines, inverter-controlled PVs and batteries. Also, for the first time, a passivity-based control is designed for two-port components of HN so that they are output strictly incrementally passive, thus finite gain stable. Examples of typical implementation are HVDC lines and FACTS.
 
(4) Assuming modular specifications of components are satisfied, we propose additional system-level feasibility conditions for subsystem HS and subsystem HN: the output of HN is in the subset of all allowed operating input space of HS. It is shown in this thesis that such a condition can be achieved by a combination of local high gain controllers and the adjustments in power output set points.
 
(5) Then, an interactive algorithm for aligning components of the EES by information exchange with neighboring components is introduced as a proof-of-concept for convergence of components to the system-level equilibrium. Such a process is the basis for the autonomous reconfigurable operation of microgrids.
 
The modular modeling and control approach introduced in this thesis is scalable.  While more work remains to fully develop this, we illustrate the possible way forward by considering the problem of enhanced automatic generation control (E-AGC) for systems with highly dynamic load variations, including effects of intermittent renewable generation.  A multi-layered yet simplified extension of the negative feedback configuration modeling is proposed for each sub-system; each subsystem interacts with the neighboring subsystems.  We show using simulations that potential instabilities between subsystems can be eliminated using distributed nonlinear control of the subsystems.