Automated robust control system design for variable speed drives

Okaeme, Nnamdi (2008) Automated robust control system design for variable speed drives. PhD thesis, University of Nottingham.

[img]
Preview
PDF
3509Kb

Abstract

Traditional PI controllers have been largely employed for the control of industrial variable speed drives due to the design ease and performance satisfaction they provide but, the problem is that these controllers do not always provide robust performance under variable loads. Existing solutions present themselves as complex control strategies that demand specialist expertise for their implementation. As a direct consequence, these factors have limited their adoption for the industrial control of drives. To counter this trend, the thesis proposes two techniques for robust control system design. The developed strategies employ particular Evolutionary Algorithms EA), which enable their simple and automated implementation. More specifically, the EA employed and tested are the Genetic Algorithms (GA), Bacterial Foraging (BF) and the novel Hybrid Bacterial Foraging, which combines specific desirable features of the GA and BF.

The first technique, aptly termed Robust Experimental Control Design, employs the above mentioned EA in an automated trial-and-error approach that involves directly testing control parameters on the experimental drive system, while it operates under variable mechanical loads, evolving towards the best possible solutions to the control problem. The second strategy, Robust Identification-based Control Design, involves a GA system identification procedure employed in automatically defining an uncertainty model for the variable mechanical loads and, through the adoption of the Frequency Domain H-infinity Method in combination with the developed EA, robust controllers for drive systems are designed. The results that highlight the effectiveness of the robust control system design techniques are presented. Performance comparisons between the design techniques and amongst the employed EA are also shown. The developed techniques possess commercially viable qualities because they elude the need for skilled expertise in their implementation and are deployed in a simple and automated fashion.

Item Type:Thesis (PhD)
Supervisors:Asher, G.M.
Summer, M.
Zanchetta, P.
Uncontrolled Keywords:Genetic Algorithms, Bacterial Foraging, Evolutionary Algorithms, Variable Speed Drives, Robust Control Systems, Programmable Load Emulators
Faculties/Schools:UK Campuses > Faculty of Engineering > School of Electrical and Electronic Engineering
ID Code:584
Deposited By:Nnamdi Okaeme
Deposited On:31 Oct 2008
Last Modified:06 Feb 2009 14:44

Repository Staff Only: item control page