Artificial Intelligence Aided Black-Box Modeling of Three-Phase Single-Stage Photovoltaic Inverter Systems

With the increasing penetration of solar energy in distribution systems, the precise modeling and appropriate control of photovoltaic (PV) generation systems are becoming increasingly significant. However, the modeling and system identification of inverter-based resources (IBRs) are challenging since sensitive information (e.g., topologies or parameters of electrical components) could not be provided by manufacturers. Black-box modeling methods that only utilize empirical data without the need for internal system details could be a useful method to solve the aforementioned issues. Meanwhile, given the strong approximation capability, artificial neural networks (ANNs) can augment the conventional modeling approaches for inverter-dominated system identification. In this paper, the black-box modeling approaches for power electronic converters (PECs) are reviewed. Furthermore, this paper proposes a data-driven black-box modeling algorithm using a nonlinear autoregressive exogenous neural network (NARX NN), aiming to estimate the dynamic behaviors of three-phase single-stage PV inverters with a hierarchical control diagram. The proposed method can predict the target output only based on the input and output measurements of the black-box system with unknown topology and parameters. Finally, simulation and experimental results are presented to demonstrate the effectiveness of the proposed work.