Set-to-set iterative learning control with applications to electrified aircraft operations

Smith, Reid Dukes

Permalink

https://hdl.handle.net/2142/129541 Copy

Description

Title

Set-to-set iterative learning control with applications to electrified aircraft operations

Author(s)

Smith, Reid Dukes

Issue Date

2025-04-24

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

Alleyne, Andrew G

Doctoral Committee Chair(s)

Alleyne, Andrew G

Committee Member(s)

• Mehta, Prashant G
• Beck, Carolyn L
• Stipanovic, Dusan M

Department of Study

Mechanical Sci & Engineering

Discipline

Mechanical Engineering

Degree Granting Institution

University of Illinois Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Iterative learning control
• linear systems
• non-repetitive disturbances
• convex polytopes
• constrained inputs
• optimal tracking time
• unattended ground sensors
• unmanned aerial vehicle

Abstract

For some iterative tracking control problems, obtaining sets at specific times is desired rather than tracking a specific reference trajectory at these times. Examples of this include a robot sorting items into a bin or a UAV flying within the communication region of ground sensors. In these problems, a control objective is to have the output trajectory lie within the sets at desired times. For any time when a set is not present, a reference composed of discrete points may or may not be present depending on the problem construction. As each set represents a region in space, multiple paths exist which satisfy this control objective due to the lack of a reference trajectory to track within each set. The sets thus increase the control design freedom as a path can be selected which lies within the sets at the desired times and minimizes other costs. Due to the repetition within iterative tracking control problems, iterative learning control (ILC) is a common control design method. While most ILC methods have not considered the presence of sets within an iteration, some methods have addressed this challenge through creating a reference within the sets. However, by creating a reference within the sets, the design flexibility which these sets provide is not used. Instead, the freedom to choose a path through these sets should be utilized within the control framework rather than restricting the path to a reference trajectory. Additionally, the presence of disturbances can lead to challenges on ensuring the output trajectory will lie within the sets at the desired time indices. This dissertation proposes a set-to-set (STS) ILC methodology to utilize the design flexibility provided by these sets. The STS ILC does not require the construction of a reference within the sets and contains guarantees on the convergence of the control input and guarantees that a feasible input exists such that the output trajectory will lie within the sets at the desired times. By updating the control law without constructing a reference within the sets, the STS ILC has guarantees on improved or equivalent performance to alternative ILC methods which do require constructing such a reference. While these guarantees are provided for the case of repetitive disturbances, unconstrained inputs, and fixed tracking time indices, the STS ILC is subsequently generalized to consider non-repetitive disturbances, constrained inputs, and tracking time indices which can be optimized. Multiple case studies are presented to demonstrate the performance of STS ILC relative to comparable ILC approaches. For the initial presentation of STS ILC and the subsequent generalizations, case studies with a linear servo system are used to evaluate the STS ILC performance and that of alternative ILC approaches. Within each of these case studies, the STS ILC obtains a lower optimization cost than the alternative ILC methods. To study the effectiveness of STS ILC for operation in a more complex environment, a case study is presented where a UAV must fly within the communication region of unattended ground sensors for each iteration of flight. Within this case study, a nonlinear plant model is used for the UAV powertrain and flight dynamics. In all case studies, the STS ILC demonstrates superior performance to other ILC methods by planning a path through the communication regions without constructing a reference trajectory within them. This results in improved performance as shown by reduced energy consumption, elimination of powertrain constraint violations, and a smoother flight path.

Graduation Semester

2025-05

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/129541

Copyright and License Information

Copyright 2025 Reid Smith

Owning Collections