报告题目：Observer design for nonlinear dynamical systems using geometrical methods
报告人：Professor Driss Boutat
Professor Driss BOUTAT received his Ph.D. degree in Differential Geometry from University of Claude Bernard, Lyon I, France, in 1993. From 1997 to 2002, he was a mathematics teacher in National Engineering School of Bourges (ENSIB). From 2002 to 2008, he was an Associate Professor in ENSIB, where he earned the Accreditation to Supervise Research (HDR) in 2007. Since 2008, he has been a Full Professor in ENSIB, which developed into the INSA Centre Val de Loire in 2014. Since 2011, Prof. Boutat is the leader of Control Team in PRISME Laboratory. He is an international expert on control and observer design. His research interests mainly focused on developing observer nonlinear dynamical systems, distributed parameter systems and fractional order systems. Until now, Prof. Boutat has published more than 100 papers in international journal and international conference. Thanks to his work, he earned the Order of Academic Palms Chevalier (Knight) since January 2010, the national award for doctoral supervision and research from 2008 to 2012, and the national award for scientific excellence since 2012. In 2011, he was selected in Who's Who in the World. Moreover, he is associate editor of Journal of Nonlinear Dynamics, member of editorial board of Discrete Dynamics in Nature and Society and member of editorial advisory board of Mediterranean Journal of Measurement and Control. In 2017, he is appointed as a foreign expert of high level by the Chinese government.
报告内容简介：For engineers, a large variety of information is not directly obtained through measurements. Some internal states are unknown or unmeasured. In order to simulate, to control or to supervise processes, and to extract important information one often has to estimate the state of the system. Online measurements of these states are often unavailable for reasons of cost or non-existence of sensor. For this, observability, observer design and inverse concepts should be considered and are important from theoretic and application points of view. A powerful response to this new challenge is the observer concept (software sensor or state estimator). This device allows estimating the state of an industrial, economic or biological system from partial measurements of the state. These measurements are obtained by mean of sensors. Thus, an observer enables to use only a minimum number of sensors to produce more information on a system. Therefore, it presents an economic interest in industry and it is important in hostile environments. This explains the intensive researches on observer design in automatic control. A simple method to observe the state of the system is to numerically calculate the successive derivatives of the measurement obtained from sensors. In practice, this method presents a disadvantage when considering noisy measurements, i.e. differentiators can amplify the influence of high frequency noise. Moreover, it is well-known that the observer based approach is a powerful concept to deal with the state estimation and is generally used in the feedback control and in the fault diagnosis problems. Observer approach is also used for estimating unknown parameters, defaults of a dynamical system.The main of this presentation is to show two geometrical methods to transform a nonlinear dynamical system endowed with an output (measurement) into the so-called observer normal form. This last enables us to use the well-known Luenberger’s observer.