AI-Based Effective Virtual Inertia Estimation in a Meshed Network of Inverters

Mass penetration of renewable energy sources (RES) has drastically triggered the integration of clusters of VSG-based grid-forming inverters (GFMIs) to enhance grid stability and facilitates the transition towards power electronics dominated grid (PEDG). Although high number of GFMIs provides a smoother frequency response due to higher emulated virtual inertia, considering the complex dynamic of the PEDG, estimating the emulated virtual inertia by the primary control layer of the GFMIs introduces extensive mathematical modeling and computational burden. To address the shortcomings of conventional model-based approaches, this paper proposes a data-driven methodology based on convolution neural network (CNN) to estimate effective virtual inertia by analyzing the explanatory frequency response of a cluster of GFMIs. The neural network (NN) is trained offline and evaluated on a sample cluster of 100% GFMIs through multiple case studies to validate the effectiveness of the proposed methodology. The application of the proposed data-driven tool can further facilitate network analysis by considering the corresponding aggregated model to enhance the stability analysis of the PEDG.