Adaptive Deep Neural Network Architecture for Data-Driven Model Based Identification of Non-Linear Dynamics of Microgrids

In microgrids, electric vehicles (EV) can act as mobile energy storage units that help balance supply and demand within the microgrid. Dynamic analysis in microgrids involves determining the distribution of active and reactive power within the system to identify power losses, load profiles of the system, and voltage profiles. Dynamic power flow analysis extends power flow studies to transient conditions and is influenced by several factors such as the state of charge (SOC) of the EV battery, grid conditions, and charging infrastructure capabilities. This work focuses on developing a data-driven model that identifies the dynamic power flow in a microgrid system using neural networks that can predict and optimize the power flow pattern in the network based on historical data. An iterative learning process is employed for model improvement by fine tuning the weights of the neural network architecture based on feedback and additional data augmentation.