Binary Neural Networks For Modular Multilevel Cascade Converters - 2023/2026

This research project, developed in collaboration between Germany and Chile, proposes the first-ever application of Binary Neural Networks (BNNs) in power electronics systems. The goal is to enhance the performance and reliability of Modular Multilevel Cascade Converters (MMCC), which are key topologies in electrical transmission and high-power applications. BNNs enable the implementation of modulation, observation, fault detection and diagnosis (FDD) algorithms, as well as fault-tolerant control, with high efficiency on real-time hardware such as FPGAs. Their advantage lies in computational simplicity, which reduces resource consumption and facilitates operation in complex systems.

The project aims to determine whether these networks can truly optimize the operation of converters such as the Dual-Star Chopper Cell (DSCC-MMCC) and the Triple-Star Bridge Cell (TSBC-MMCC). Expected outcomes point to greater robustness, efficiency, and responsiveness in modular converters, opening a new field of artificial intelligence applications in power electronics.

Program: DFG Germany

Project Duration: 36 months

Principal Investigator: Javier Pereda Torres