Data-driven identification of power converters from HIL simulations
Prof. Bertrand Cornélusse
University of Liège, Belgium
This lecture will describe how to perform experiments to collect data from a power converter using a HIL simulator, in order to build a multi-model of the power converter that can be integrated into a power system simulation tool. Based on our recent research papers (Interpretable large-signal black-box models of power electronic converters: a multi-model approach and A C-HIL-based data-driven DC–DC power electronics converter model for system-level studies), we will first review the identification procedure for the linear models, then demonstrate how to combine them within the proposed neural-network-based multi-model structure, and finally show how to integrate this model into a power system simulator.
Biography
Bertrand Cornélusse received the M.S. in electrical engineering and Ph.D. in engineering sciences degrees from University of Liège, Belgium, in 2006 and 2010, respectively. His interests lie in the application of optimization and machine learning to energy management applications, such as microgrids, distribution systems and electricity markets. He is now professor in “smart microgrids” within the Electrical engineering and Computer Science department of the University of Liège. Before this he has led the development of the European day-ahead market coupling algorithm Euphemia together with European exchanges, and contributed to several European research projects and research projects funded by the Walloon region of Belgium, such as the GREDOR project that studied the application of optimization and machine learning techniques to improve the planning and operation of electrical distribution systems.