Neural Control and Monitoring for Modular Storage to Storage DCDC Conversion

In this work, a strategy for the control of a modular multiport DC-DC converter is proposed. The converter is obtained by the extension of the classic split-pi topology considering one input port with a PV source and three bidirectional ports with generic electrical storage systems. The control strategy employs an Artificial Neural Network (ANN) for feed forward control to achieve faster closed loop response. The ANN implements a secondary service output for state of charge estimation to monitor the charge and discharge process of the storage element. The system is implemented in PLECS environment for both training dataset acquisition and validation in time-domain simulation.