Abstract
This paper presents a simple modification which can be made to a pre-designed dynamic anti-windup scheme in order to improve its performance. Roughly speaking, the modification enables the dynamic anti-windup compensator to act more like a static anti-windup compensator in certain circumstances. In particular, the modification enables the output of the compensator to decay more quickly than if it were absent, thereby effecting a swifter recovery of linear behaviour. The modification is therefore suitable for – and indeed motivated by – applications where the original anti-windup compensator contains slow poles, resulting in a potentially lengthy recovery of linear behaviour. The paper describes in detail the modification and presents conditions under which it is able to preserve stability.
Highlights
Anti-windup compensators supplement baseline control systems in order to improve their performance during control signal saturation
This paper proposes a nonlinear modification which can be retro-fitted to a predesigned anti-windup compensator to enable the output of the anti-windup compensator to decay more rapidly after saturation has ceased
This paper has proposed a simple nonlinear modification filter which can be appended to a linear dynamic anti-windup compensator in order, potentially, to improve performance of the overall system
Summary
Anti-windup compensators supplement baseline control systems in order to improve their performance during control signal saturation. The nonlinear modification operates by monitoring the level of saturation, and attenuating the magnitude of the anti-windup compensator’s output depending on this: the nonlinear modification can be considered as a nonlinear gain It transpires that the use of this nonlinear gain, together with a simple filter of order equal to the number of anti-windup outputs, can enable stability to be maintained whilst improving small signal performance. These ideas were motivated by a flight control application considered by the authors ([8,9] and in unpublished work) which suffered a similar issue to that described above. A positive (negative) definite symmetric matrix M is indicated by M > (
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
