Wided Belloumi defends her PhD on Nov. 29, 2021 at 1:30 PM.
Place : Ecole Centrale de Lyon, Amphi 202
Jury :
Rapporteurs :
Ferid Kourda,Université Tunis El Manar; Edith Clavel, G2ELab
Autres membres :
François Costa, SATIE; Ferid Kourda, Université Tunis El Manar; Fethi Couhoubani, Université de Tunis Carthage; Jaleleddine Ben Hadj Slama, LATIS-Laboratory, Ecole Nationale d’Ingénieurs de Sousse
Encadrement :
Arnaud Bréard et Christian Vollaire, Ampère
Abstract :
With the current evolution of electronics, the development of power electronic systems takes a very important place in various applications such as automotive and aeronautical applications. Power converting devices are rapidly developing to achieve high efficiency and high power. One of the downsides to the increased use of electronic devices is that they must operate at high switching frequencies, which creates compatibility problems. It is therefore necessary to avoid that the device developed does not produce too much disturbance in its environment. The discipline of electromagnetic compatibility has been incorporated so far at the end of the development phase of a system, before the stage of certification and industrialization. Taking these constraints into account from the design phase can minimize costs and broaden the choice of available solutions at lower cost by reducing the testing phases. In this context, this thesis proposes an automatic optimization approach to the design of a converter in view of EMC constraints. The first step of this thesis is the design of an automatic printed circuit using a Dijkstra algorithm coupled with the genetic algorithm. The second step is to put in place a modeling approach compatible with the optimization process. It is then essential to take into account all the parasitic elements, namely the elements in the RSIL, the converter in order to properly estimate the conducted disturbances such as the common mode and differential mode voltages. Once the emission models have been established, an optimization process is proposed. This approach achieves the optimum converter geometry with the lowest parasitic effect to meet EMC requirements. Finally, the results obtained thanks to the genetic algorithm used are presented and studied. The effectiveness of these results has been validated by experimental results. Our contribution consists in developing an automatic optimization approach allowing obtaining converters having an optimal electromagnetic behavior before moving to the filtering solution which will reduce the cost and the volume of the filter.
Keywords :
Electromagnetic compatibility, Conducted emissions, Modeling, Dijkstra algorithm, Genetic algorithm, PCB layout, Automatic optimization