Tandem cold rolling presents difficult challenges to the control engineer: the mill is a nonlinear multivariable system, limited instrumentation is available to measure product quality, the deformation process is uncertain, there are significant material transport delays and practical mill operation is required over a wide range of speeds with a variety of rolled products, introducing significant variation in the mill dynamics. A robust multivariable control solution is sought in this thesis. A nonlinear simulation of an open-loop mill is developed and validated against published data. A methodology for designing H loop shaping controllers is presented. Controller development, based on a typical industrial structure, is undertaken for comparative purposes and both types of controller are evaluated at full mill speed in the face of a wide range of disturbances. A gain scheduling scheme, with rules for weight modification at the intermediate operating points, is designed for mill acceleration. Potential for improved product quality is seen with the H controller. Theoretical studies into H optimization problems are also undertaken. In particular, the well-known problem of pole-zero cancellation is addressed. A novel construction of weighting function in a mixed sensitivity design enables the cancellation phenomenon to be exploited and is shown to act as a mechanism for partial pole placement. It is also shown how pole-zero cancellation arises with H loop shaping for particular classes of plant. Further perceived shortcomings of the H loop shaping method, namely the relationship between shaped and nominal plant uncertainty and the mathematical intractability of the uncertainty model, are afforded consideration. Several structured uncertainty models are investigated and relationships uncovered between H loop shaping and -synthesis. A structured loop shaping design is motivated: a solution via - K iteration yields an interesting robust stability interpretation, while the design is shown to have attractive robust stability and robust performance potential.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:696474 |
| Date | January 1998 |
| Creators | Geddes, Eric John Muir |
| Publisher | University of Leicester |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/2381/30171 |
Page generated in 0.0198 seconds