Black-box model for estimating efficiency curves in DC-DC converters for energy storage systems

The objective of this study was to develop a black-box model capable of estimating the efficiency of different DC-DC power converters based only on their label data (nominal voltage, power, peak efficiency…) and operating conditions. This was done by measuring the efficiency at several operating points of five converters designed by Lyra Electronics for energy storage systems, and using the collected data to train an Artificial Neural Network (ANN). Moreover, the ideal number and combination of predictor variables were determined using both the Stepwise Selection (SS) and the Backward Stepwise Selection (BSS) methods. After comparing the results, it appears that the latter method generates slightly less accurate models, but in much shorter time. In any case, the models show very high accuracy, making this black-box technique a promising tool for high-level energy systems design. However, it is necessary to test this methodology on a larger and more varied set of efficiency data to better assess its validity.