Professor, University of Strasbourg
Dominique KNITTEL received the Engineer diploma from the Ecole Nationale Supérieure de Physique de Strasbourg, France, in 1987 and the Ph.D degree from the University of Strasbourg in 1990, both in Control Engineering. Since 1992 he has been at University of Strasbourg where he is currently full Professor in mechatronics.
His research interests include modeling, robust control, optimization, 1D-3D simulations, mainly for manufacturing systems, industrial Roll-to-Roll systems and renewable energy. He has directed more than thirty PhD and Master thesis. He received the ISA Transactions best Paper Award 2012 (Journal - Elsevier) and the Excellent Paper Award at the IFIP-ICSI 2013 conference
Enrico Nobile will chair the Academic Roundtable that will see professors from distinguished universities around the world discussing forefront optimization-related research projects and applications. Participants can contribute with their views on the challenges of these research activities and cast a light on the skills and knowledge that future designers must-have.
In today industrial context, constraints on manufacturing process are constantly tighter. In order to meet these constraints, the control optimization cannot longer rely only on traditional synthesis tools. Genetic algorithms, useful for complex non-convex problem optimization, are used in this work.
The most intuitive way to optimize the performances of a mechatronic system is to proceed with time domain simulations and to reduce, for instance, the error between measured outputs and references. Another common objective is to maximize the robustness of the controlled plant (closed-loop) to parameter variations by evaluating cost functions for many parameters values. However, time domain simulations are time consuming and the results are depending on the system inputs.
In this work, the multi-objective optimization is made in the time and frequency domain and differences are evaluated. Moreover, a new optimization strategy, combining time and frequency domain and using MORDO approach is proposed. This work set up a design methodology which considers the electro-mechanical system to be designed as a whole and no longer as the sum of its parts.