Discrete Optimization of EMI Filter Using a Genetic Algorithm
Résumé
A discrete optimization using a genetic algorithm (GA) for electromagnetic interference (EMI) filer design is presented in this paper. Using a macro-model in the frequency domain this approach takes into account the load impedance and the source emissions. The macro-model includes an accurate filter model that takes into account the PCB stray elements and the passive components high frequency complex behavior. A framework between on two software is established. This framework permits a data exchange between the filter accurate model and the GA algorithm. This approach is applied on a Class-D ampliftcation system for validation. Using the suppliers well-defined surface mounted devices (SMD) database, the GA search for a set of components that respects the Electromagnetic Compatibility limits whlle reducing the filter power losses.