Dr J Anthony Rossiter
Reader
Address:
Dr Anthony Rossiter, MA, DPhil, MIEE, CEng
Department of Automatic Control and Systems Engineering
University of Sheffield
Sheffield
S1 3JD
Tel: (+44) (0)114 222 5685
Fax: (+44) (0)114 222 5683
Email: j.a.rossiter @ sheffield.ac.uk
Room: B12, Amy Johnson Building
Current research interests
Professional activities
Publications and Summaries of selected recent publications
Awards
Collaborations
Current Teaching
Current Administration
Career History
Current research interests
With collaborators, I have made numerous contributions within the literature, perhaps the most well known being related to the closed-loop paradigm for predictive control, a concept that is now very widely adopted and often referred to using terms like terminal mode or terminal weights. I continue to have an interest in the development of this area have given numerous invited presentations where the personal flavour I give is very much linked to simplicity.
Guaranteed stability of predictive controllers
Conventional predictive controllers are not guaranteed to be stabilising (although they usually are). Dr Rossiter was actively involved in the solutions generated by academia to this issue and proposed mechanisms for ensuring apriori stability with only mild assumptions. Usually these algorithms outperform GPC and are of similar complexity. Dr Rossiter's particular contributions are focused on the advantages of the closed-loop paradigm for predictive control. In this the constrained optimisation over future control moves is replaced by a fixed controller and perturbations to the loop input. These approaches are algebraically equivalent but the latter gives numerical advantages and can also improve robustness.
Feasibility of predictive control in the presence of constraints.
Predictive control allows optimisation of predicted performance subject to constraint satisfaction. Careless set up of the problem can lead to a sequence of optimal inputs selected at the current sample being poor or even unavailable at the next sample. When one is unable to satisfy constraints, this is called infeasibility. Usually feasibility is ensured by a supervisor making sensible demands of the control law. Dr Rossiter's work has concentrated on how the control law itself can prevent self generated infeasibility. This is particular important where there is a combination of hard output constraints and unknown disturbances.
Computational and algorithmic simplicity while constraint handling
Typical predictive controllers use quadratic programming to find the optimal control. There are case however where this size of problem may be intractable, or with the complexity of such a solution it is hard for plant operators to gain confidence in. Dr Rossiter has proposed several suboptimal alternatives to the QP optimisation which reduce the online computation to a set of simple inequality checks and also relate the decision making close to the intuitive understanding of unit operators. He is currently investigating and developing the industrial algorithms of IDCOM and PFC which are 'simpler' MPC algorithms than commonly adopted in academia. He is also looking at how to simplify the parametric algorithms that have been popularised recently.
Non-square systems
Non-square systems are highly prevalent within industry, but in fact are little studied within the academic literature. A recent project, currently led by Leo Shead, is looking at aspects of the control of non-square systems, especially where they link to predictive control. We expect to begin publishing some interesting work in this are from late 2007.
Controller robustification
Dr Rossiter has shown the links between Youla parameterisations and the conventional predictive control structure. Hence, a Youla parameter can be incorporated into the predictive control design to reduce the sensitivity functions of one's choice in a systematic fashion, without loss of nominal performance. Recently with collaborators he has extended this approach to make use of invariant ellipsoidal sets which are a convenient tool for handling model uncertainty. The main contribution is to increase the volume of the robustly stabilisable regions without incurring a large online computational burden. A project of particular interest is how much robustness one can get while constraint handling with a limited online computational burden.
Education
Engineering education in the UK is undergoing rapid change to keep up with the changes in school curriculum and large increases in numbers which is introducing a larger variability of backgrounds. One issue of particular importance is how to change teaching styles, content and support mechanisms to help these students both to progress and enjoy their time at university. Partly, I pursue this through my role as the Faculty DLTD, but I also lead some individual projects and have published several conference papers.
I have proposed and obtained internal funding for several projects (some with colleagues) to enhance teaching within my department and the faculty. This include innovative use of the virtual learning environments and team teaching.
A current project is looking at how to enhance the mathematics learning support within the faculty and the University.
As a control engineer, I am also leading a HEA sponsored special interest group (2007) which has the aim of encouraging control engineering colleagues to share quality teaching resources more effectively. I am also a member of the IFAC technical committee on control education (EDCOM).
Miscellaneous other interests:
Dr Rossiter has interests in a number of other fields such as superheater steam temperature control, modelling for predictive control and multi-rate systems, where he has published. Current PhD students are looking at areas such as fault detection and collision avoidance within both aerospace scenarios and for road traffic.
Professional activities
I am an external examiner for Reading University (2009-12) and Portsmouth University (2011-14).
I act as an editor of the Journal of Adaptive control and signal processing and a reviewer for numerous journals and conferences.
As a member of the UK control chapter for the IEEE I have organised several colloquia (2005-2008).
I also have a rapidly developing interest in engineering education for which I have received a major award.
I am General chair of Advances in Control Education August 2013 (details to be confirmed).
I am co-ordinator of a UK special interest group and serve on the IFAC education committee (EDCOM) and I will be involved in IFACs pending initiative in developing a repository of control resources. More information is available at:
http://controleducation.group.shef.ac.uk/
Publications and Summaries of selected recent publications
A complete list of my publications is available from the link above right and a shorter list of recent reports together with full text download is available from the following link:
Summaries of selected recent publications are shown in the table below:
| Details of Publication | Brief description of contribution | |
|---|---|---|
| 1 | Interpretations of and options in dual-rate predictive control, J.A. Rossiter, J. Sheng, T. Chen and S.L. Shah, 2005, Journal of Process Control, 15, 135-148 | A bit like a survey paper in that it gives a critique of many existing proposals and then proposes some simple strategies that give you better performance and often easier implementation. |
| 2 | Using interpolation to improve efficiency of multiparametric predictive control, J.A. Rossiter and P. Grieder, 2005, Automatica, 41, 4 | Shows how interpolation can bring about an order of magnitude reduction in the complexity of parametric solutions, often with negligible loss of performance and no loss of feasibility. |
| 3 | Stable prediction for unstable independent models, J.A. Rossiter, IEEE Trans AC, 2003, 48, 11, pp2029-2035 | Fills an important gap in the literature for predictive functional control which previously used rather ad hoc means of coping with open-loop unstable systems. This paper gives a numerically robust method, that also facilitates better closed-loop performance. |
| 4 | Who needs QP for linear MPC anyway, B. Kouvaritakis, M. Cannon and J.A. Rossiter, 2002Automatica, 38 (pp. 879-884). | Gives a high profile to alternative methods for finding a constrained optimal. Shows that if one uses ellipsoidal invariant sets, then one can pose a linear constrained optimisation as a simple line search and therefore with very fast convergence. This was a novel insight in the context. |
| 5 | A data driven subspace approach to predictive controller design, R. Kadali, B. Huang and J.A. Rossiter, Control Engineering Practice, 2003, 11, 3, 261-278 | Modelling is a key part of predictive control that is often neglected in the literature. This paper investigates the potential of a an emerging modelling technique and shows how the synergy with predictive control objectives can facilitate much more effective control design. Also derives some important technical details. |
| 6 | Interpolation based computationally efficient predictive control, J.A. Rossiter, B. Kouvaritakis, M. Bacic, 2004, IJC, 77,3, 290-301 | Gives a survey of recent interpolation techniques with a critique of the range of application, performance and implied computational load. A good start point for future developments. |
| 7 | I. S. Choi, J.A. Rossiter and P. J. Fleming, Looper and Tension Control in Hot Rolling Mills: a survey, Journal Process Control, to appear 2007 | Gives a thorough survery of different control techniques that have been used to control hot rolling mills, focussing on strengths and weaknesses. Includes proposals for development including illustrations of some predictive control applications. |
Awards
University of Sheffield Senate award for sustained excellence in learning and teaching, 2009
(See: http://www.shef.ac.uk/lte/reward/winners08.html)
The Higher Education Academy Engineering Subject Centre Teaching Award 2008.
(See: http://www.engsc.ac.uk/an/teaching_awards/awards07_08.asp)
Engineering Teaching prize from Royal academy of Engineering, 2005.
(See: http://www.raeng.org.uk/prizes/etp/default.htm)
Outstanding reviewer for Automatica, 2004-05.
HEAVISIDE PREMIUM awarded (Oct 28 1998) by the Institution of Electrical Engineers for: Constrained cautious stable predictive control, J.A. Rossiter, J.R. Gossner and B.Kouvaritakis, 1997, Proc. IEE Pt.D, 144, 4, pp313-323
Best paper award ISSC, Maynooth, Ireland, 2001, Extending the stability guarantees for an efficient predictive control law, J.A. Rossiter, pp84-89
Member of IEEE Control chapter which received outstanding chapter award in 2006.
Collaborations
Dr Rossiter enjoys collaborating with many people as and when the opportunities arise. He has a long standing collaboration with with Prof. Kouvaritakis (Oxford) and Dr M Cannon (Oxford University). Moreover he has also published joint works with, for example, researchers such as Professor L Chisci (Firenze, Italy), Dr P Neal (Powergen), Prof SL Shah and T Chen (University Alberta, Edmonton), Prof J Richalet (Adersa), Dr Shead (Surrey) and Prof Muske (Villanova, USA) and many Ph.D students from other institutions than his own.
Current Teaching
- ACS108 Laboratory and Professional Skills
- ACS124 Modelling, Simulation and Control
- ACS123 Systems Engineering Mathematics
- ACS221 Control Systems Analysis and Design
- ACS6012 Predictive control
Current Administration
Academic leader for Maths and Statistics Help (MASH).
Chair of Department Learning and Teaching Committee and co-leader of 1st year teaching team.
I have developed and lead numerous departmental activities that seek to engage and enthuse school age pupils in engineering.
Career History
1984-87 MA 1st class, Engineering Science, Oxford University
1987-90 D Phil 'Multivariable self-tuning', Oxford University
5/90 - 9/92 EPSRC Research Assistant Oxford University
10/92 - 2/01 Lecturer, Mathematical Sciences Loughborough University
Promoted to Senior Lecturer in 1998
3/01 Joined AC&SE at The University of Sheffield
Promoted to Reader (10/01)
Also I am a chartered engineer and a HEA registered practitioner. In my spare time I attend church, manage children´s football, play and referee, and help out with fundraising for a local school.