Seminars Summer 2007

Le Laboratoire d’Automatique de l’EPFL a le plaisir de vous inviter aux séminaires selon la liste ci-après. Une mise à jour régulière des informations concernant ces séminaires est disponible à l’adresse sur cette page. En particulier, il est conseillé aux visiteurs externes de vérifier que les séminaires soient dispensés comme prévu ci-dessous.

Where: Salle de séminaire LA-EPFL, ME C2 405 (2è étage), 1015 Lausanne

When: Friday at 10h15


Summer 2007 seminars

Implantations Industrielles de la Commande Prédictive PFC

23.03.2007 – Dr. J. Richalet – Consultant-Formateur.

Après une présentation rapide de l’origine Genevoise de la Commande Prédictive, on expose les 4 principes fondamentaux de cette commande qui s’est développée à partir de la fin des années 1960. Les premières applications étaient dans l’industrie pétrolière et dans les systèmes d’armes, mais cette technique est aujourd’hui largement diffusée dans de nombreux pays et pratiquement dans tous les secteurs industriels. Sa large diffusion vient de ses objectifs: être dans la continuité de la commande industrielle PID classique, simple de compréhension, d’implantation et de réglage. Elle est maintenant de fait implantée dans tous les automates de l’industrie de production, avec des temps de retour courts. On citera donc des applications industrielles diverses de la commande PFC (Predictive Functional Control), en particulier celles qui concernent les industries où le phénomène physique principal est “l’échange thermique”: chimie, pharmacie, alimentaire, climatisation, fours, énergie, etc.

Adaptive Vibration Compensation at the ESO VLT Interferometer

30.03.2007 – Dr. N. Di Lieto – European Southern Observatory (ESO), Munich, Germany.

Fringe tracking, also known as co-phasing, is a method of stabilizing the Optical Path length Difference (OPD) from the observed astronomical source to the instrument detector via the different telescopes in an interferometric array. The objective is to stabilize the OPD to within a fraction of the observing wavelength (goal: 100 nm rms), so as to maximize the integration time of scientific instruments. In addition to OPD, due to large but predictable geometric changes and random but relatively slow atmospheric turbulence, vibrations generated by the telescope infrastructure and propagated to the optics with an amplitude of 1 micron and frequencies up to 100 Hz must be compensated in real time. This seminar introduces an adaptive vibration compensation method, which is currently in use at the ESO VLT Interferometer.

Optimal Process Control

04.05.2007 – Prof. M. Fikar – Institute of Information Engineering, Automation, and Mathematics, Slovak Uni-versity of Technology (STUBA), Bratislava, Slovak Republic.

In this lecture, we give an overview on numerical approaches to optimal control of processes. It will be shown how dynamic optimization problems can be transformed into nonlinear programming problems using various transformation schemes. Selected problems and usage of dynamic optimization packages will be illustrated on optimal control and estimation of chemical and biochemical processes. In the second part, connections between dynamic optimization and model-based predictive control will be explained, and various problems in explicit approach to predictive control will be studied.

Estimation de l’État de Fonctionnement de Systèmes

11.05.2007 – Prof. J. Ragot – Centre de Recherche en Automatique de Nancy, Institut National Polytechnique de Lorraine (INPL), Nancy, France.

La problématique de la sûreté de fonctionnement se décompose en deux sous-problèmes. Le premier s’applique, dès la conception d’un système, à garantir qu’il fonctionnera selon un cahier des charges donné, même si une partie des fonctions n’est plus assurée. Le second sous-problème, lié à la phase d’exploitation, est dédié à la surveillance en ligne pour détecter un défaut, le localiser, combattre ses effets en mettant en oeuvre une commande qui le tolère. Au cours de l’exposé, on insistera particulièrement sur la génération d’indicateurs de défaillance et la structuration de ces indicateurs pour localiser les défauts. Quelques applications traitées au CRAN seront présentées. Quelques problèmes difficiles seront exposés (détection de changement de mode de fonctionnement, diagnostic de systèmes décrits par des multi-modèles, prise en compte des incertitudes des modèles et des mesures).

Systems Bio(techno)logy: The Process Systems Engineering of Cellular Processes

01.06.2007 – Prof. V. Hatzimanikatis – Laboratory of Computational Systems Biotechnology (LCSB), EPFL.

Biological sciences and every field in biotechnology are experiencing the on-going revolution of systems bio(techno)logy. It is an information revolution driven by advances in analytical technology, biochemistry, nanotechnology, polymer chemistry, and material science. These technologies enable the precise and quantitative characterization of the various molecules within a cell and the monitoring of many cellular processes simultaneously. Systems bio(techno)logy offers two main opportunities to process systems engineers: (i) contribution to technology development, and (ii) meaningful analysis of the large-scale information generated by these technologies.
We will offer a perspective on the challenges posed to process systems engineers in the development and implementation of computational tools and frameworks for the management, analysis and interpretation of biological information, and for the integration of analytical technologies and computational tools for the elucidation and (re)engineering of the function of cellular processes. We will outline the capabilities and the limitations of some of the current technologies used in studying cellular function and discuss qualitative similarities between problems in process systems engineering and systems bio(techno)logy.

Control Challenges in Diabetes

15.06.2007 – Dr. T. Prud’homme – Disetronic Medical Systems AG, Burgdorf, Switzerland.

In type 1 diabetes, the pancreas stops producing insulin, and this deficiency must be compensated by an exogenous delivery of insulin. Ideally, an insulin pump that adjusts continuously the infusion rate could do this. The insulin profile should be optimized so as to maintain the blood glucose concentration within a reasonable range. From a control perspective, the insulin infusion rate represents the manipulated variable and the blood glucose concentration the controlled variable. It is important to stress that blood glucose concentration is measured about 4 or 5 times a day, which precludes the use of a fully closed-loop controller. Thus, the use of model-based “weakly” closed algorithms has become very attractive. However, the relationship between infused insulin and glucose concentration is not fully understood. It has been shown that numerous factors have a significant influence. A precise quantification of these factors is tedious because of the lack of understanding of the underlying phenomena, along with the lack of measurements that would make it possible to decouple these effects. In this talk, we describe the aforementioned control challenges and present on-going research efforts in this direction at Disetronic-Roche.