Decentralized AI-based Fault Detection and Localization to Enhance Dynamic Response of Grid-Forming Inverters

Grid-forming inverters (GFMIs) are promising solutions for voltage and frequency support in upcoming power electronics-dominated grids (PEDG). However, current state-of-the-art decentralized control schemes for GFMIs are designed for operation under normal conditions. These decentralized control schemes could result in an adverse dynamic response if a cluster of GFMIs network disconnects from the rest of the grid due to a fault. The dynamic response of GFMIs could be improved with coordinated control schemes which require communication and make the system more complex. This paper proposes a decentralized Artificial Intelligence (AI)-based method for online fault detection and localization to enhance the dynamic response of GFMIs without using a communication layer. Each GFMI will detect and localize the line tripping based on its output active and reactive power measurement. Then, according to the topology of the grid after line tripping, each GFMI updates its nominal power and inertia to suppress the frequency transient caused by the line tripping. The inertia and damping factor of the GFMIs are then recalculated to ensure optimal operation of the network of inverters after line tripping. Several case studies are presented to validate the effectiveness of the proposed method in the timely detection, localization, and mitigation of adverse frequency transient after line tripping in a communication-less manner.