Accommodating magnitude and rate-limits in adaptive control systems Seminar
- Time:
- 16:00 - 17:00
- Date:
- 16 March 2023
- Venue:
- Building 25, room 2015
For more information regarding this seminar, please telephone Vanui Mardanyan on isvr@southampton.ac.uk .
Event details
ISVR Research seminar
All real control systems have to take account of the fact that the control signal (the manipulated variable) will be limited in magnitude and rate. These limits are not accommodated easily in standard linear, frequency-domain controller design. However, a lot of progress in so-called anti-windup design has occurred over the last 20 years so that standard (linear) control systems are now straightforward to enhance with anti-windup compensators guaranteeing rigorous stability and performance guarantees. Adaptive control is an inherently nonlinear design method and magnitude and rate-limits are particularly pernicious in these control systems because they lead not only to standard "windup" like effects but also corrupt the adaptation law. In this talk, anti-windup-like remedies will be proposed for adaptive controllers and it will be shown that, under reasonable assumptions on the underlying systems and references that these remedies enable i) stability of the overall closed-loop to be maintained and ii) the performance of constrained model-reference adaptive control (favoured by Boeing in their advanced systems ) to, in a sense, approach those of unconstrained MRAC systems.
Speaker information
Prof Matthew Turner,is a professor in ECS, having joined in 2020 after spending more than 15 years as an academic at University of Leicester where he became head of the Aerospace and Computational Engineering Research Group. His research interests are broadly in robust control, nonlinear control and applications mainly involving aerospace systems. He received funding for 8 years from MBDA for his work on anti-windup, robust control and adaptive control and was involved in a series of flight-test campaigns, with DLR Germany, designing and implementing anti-windup compensators for the alleviation of pilot-induced-oscillations. He has also applied these techniques to robotic telescopes, drive-by-wire systems and various autonomous systems. He is currently interested in generating theoretically sound and systematic approaches for blending standard and learning control systems.