Optimal design of unmodeled linear systems using control-based continuation

Cui, Tonghui

Permalink

https://hdl.handle.net/2142/97472 Copy

Description

Title

Optimal design of unmodeled linear systems using control-based continuation

Author(s)

Cui, Tonghui

Issue Date

2017-04-26

Director of Research (if dissertation) or Advisor (if thesis)

Dankowicz, Harry

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

M.S.

Degree Level

Thesis

Keyword(s)

• Control-based continuation
• Model-free
• Optimization

Abstract

This thesis describes the use of control-based continuation for design optimization, in the presence of constraints and without access to a model, of the response of a linear system to harmonic input. A proof of concept of this paradigm is presented in the context of an armature-controlled DC motor. Specifically, three design problems are formulated with the objective function equal to the maximum angular velocity response to a harmonic torque disturbance, and a constraint that is imposed on each of three distinct stability margins, respectively. The analysis shows that the simulation model for the DC motor may be treated analogously to an actual experiment with all information drawn from real-time measurements, rather than from the model itself. The control-based continuation paradigm is formulated in terms of a non-invasive, yet locally stabilizing control scheme, which can be tuned to accelerate convergence to the steady state response. The numerical analysis uses the matlab-compatible continuation platform coco to determine the implicit relationship between model parameters that results from the constraint, and to evaluate the objective function along the corresponding constraint manifold. A comparison between a scheme that relies on finite differences for approximating the problem Jacobian and an algorithm based on the Broyden update is also included.

Graduation Semester

2017-05

Type of Resource

text

Permalink

http://hdl.handle.net/2142/97472

Copyright and License Information

Copyright 2017 Tonghui Cui

Owning Collections