Enhancing Artificial Intelligence-Enabled Power System Stability Assessment Through Digital Twin-Based Testbeds

The increasing share of inverter-based resources (IBR) creates stability challenges for modern power systems. Artificial intelligence (AI)-enabled methods have been developed to assess power system stability following contingencies. The development and validation of AI algorithms are usually carried out offline, using available datasets. Advancements in AI-based solutions make it possible to analyse the growing volume of data acquired from power systems. A digital twin (DT) is a virtual replica, continuously reflecting the behaviours of physical systems. This paper utilises a DT-based testbed that provides a realistic and interactive environment for the development, testing, and validation of AI algorithms to guarantee their reliability and effectiveness. A case study of an AI-enhanced real-time transient stability assessment (RT-TSA) is performed to demonstrate the utilisation and value of the DT-based testbed. Multiple algorithms are assessed through the DT, enabling the selection of the most suitable one based on offline accuracy and online execution time. The use of DT-based testbed has significant potential to advance AI-based methods across various power system applications.