Approche système pour l’étude de la compatibilité électromagnétique des réseaux embarqués

摘要

Energy saving in stationary or embedded systems is a general trend in the modern society. Therefore, the “More Electrical” concept is widely developed, using the Power by Wire idea. The need of increased efficiency and the various ways the electricity is produced, used or stored has led to the generalization of power electronics use. If this solution is effective regarding weight and losses, the high switching frequencies and sharp commutations (several mega-volts or mega-amps per microseconds) generate Electromagnetic Interferences (EMI) which have to be managed. This phenomenon is especially dramatic with the new wide band gap devices, with always increasing commutation speed. Electromagnetic models (EMC) of power electronics converters are thus needed to manage EMC aspects of More Electrical Systems. Depending on the needs, many solutions have been proposed in the literature, to account for the high frequency behavior of power electronics converters (from the switching frequency to several tens of Megahertz). In addition to the classical normative approach, the “EMC system model for power electronics converter” presented here aims to be suitable for embedded networks. In opposition to EMC filter design method, no inner knowledge (as an accurate description of each element and propagation path) from the studied converter is needed. Only external measurements are needed to get the model. Thus, non-disclosure agreement is guaranteed and the embedded network can be studied. Regarding the network structure, the “LISN + Converter” approach can be far away from its complexity. A more global approach might be achievable with “black-box” approach. For the normative approach, only EMI under 30MHz are considered. By increasing the switching frequency, the “EMC system model” has to be valid up to 100MHz. The aim of the Ph.D. is to achieve an entire identification protocol of a “Black-box” model. This modification has been chosen for: • Its tiny number of elements. This means that it can be use in network analysis with multiple converters. • Its generalist form lead to a systematic method of analysis. • Its links with the classical common mode and differential mode approach which give some interesting connection with classical converters modelization. Those links lead to a physical consideration about the meaning of this non-comprehensive model.