Abstract
dc:description.abstractIterative learning control (ILC) is a control method that is well-suited for repetitive tasks with finite durations. ILC algorithms learn from the past input and error information to improve the system's tracking performance. In particular, it can achieve perfect tracking even in the presence of model uncertainties. Starting from the work by \cite{arimoto1984bettering}, ILC has attracted intensive research in the last few decades and now has many applications.<br/><br/>Depending on whether the system model information is utilised, ILC designs can be classified as model-based ILC and model-free ILC. With system model information, model-based designs can achieve excellent tracking performance, e.g. monotonic convergence. In practice, the accurate system model may be expensive to obtain, making model-based algorithms challenging to implement. To address this limitation, this thesis studies data-driven ILC algorithms using the well-known Willems' fundamental lemma, which shows that linear system behaviour can be presented by a given persistently exciting trajectory.<br/><br/>Based on Willems' fundamental lemma, a novel data-driven norm optimal ILC (NOILC) algorithm is proposed to eliminate the requirement of the model-based NOILC algorithm on the system model. It shows that the developed algorithm achieves identical tracking performance to the model-based NOILC algorithm. We also show the algorithm can handle system constraints. Two extensions are then developed to relax the persistent excitation condition of the existing input data: the receding-horizon-based data-driven NOILC algorithm and the trial-partition-based data-driven NOILC algorithm. To demonstrate the generality of our proposed data-driven framework, we study two practical ILC applications, namely point-to-point tracking tasks and consensus tracking tasks. The simulation examples are provided to verify the effectiveness of the proposed algorithms.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jiang, Zheng
- Advisors dc:contributor.advisor
-
- Chu, Bing
- Rapisarda, Paolo
- ORCID person:identifier.orcid
Rights
- Language dc:language
- English, eng